Bone Formation During Distraction Osteogenesis Is Dependent on Both VEGFR1 and VEGFR2 Signaling

Materials

Male C57B6 mice (9–12 wk old) were purchased from Jackson Laboratories (Bar Harbor, ME, USA). DO devices were obtained from KLS Martin L.P. (Jacksonville, FL, USA). Both anti-VEGF antibodies, MF-1 (anti-VEGFR1) and DC101 (anti-VEGFR2), were obtained from ImClone through a materials transfer agreement. All reagents for the PCR analysis were from Applied Biosystems(Foster City, CA, USA), and plate assays were read on an ABI 7700 Sequence Detector (Applied Biosystems, Foster City, CA, USA). All external and internal primers with the exception of that used for osteocalcin were from commercial sets available from Applied Biosystems.

Experimental design

All animal research was performed under an approved IACUC protocol. Animals were randomly distributed into 1 of 16 groups (5 control time points, 5 single antibody, and 5 double antibody experimental time points, and 1 reference that was the unoperated mid-diaphyseal tibia bone at day 0). Time points of the study were day 7, which is the end of the latency period; day 10, which is 3 days of active distraction; day 17, which is 10 days of active distraction and is the last day of distraction; day 20, which is 3 days into the period of consolidation; and day 31, which is 14 days into the period of consolidation and is the end of the experiment. In the antibody blockade groups, the antibody administration began at 3 days after surgery, and administration was continued thereafter every 3 days. A total of at least 129 mice were used to complete the study.

μCT evaluation for vessels and bone tissue formation was performed on separate sets of samples. The group size for both μCT experiments was N = 3, and a total of 18 mice were used in aggregate for the μCT experiments. Vessel evaluation was performed at 17 days after surgery, and bone evaluation was performed at 31 days after surgery. The group size for histological assessments was N = 5 (15 mice total). Comparative analysis between the three experimental groups was performed on 31-day postsurgery specimens. An additional group of 21-day postsurgery control (no treatment) specimens was included in the immunohistological analysis. The mRNA assessments were carried out on three separate pools of total RNA, with each pool prepared from N = 2 or 3 mice for each time point. At least 30 controls, 30 single-, and 30 double antibody–treated animals were used for these experiments. Pooled sample sets from the unoperated bones were used as the reference, which was prepared from at least six animals.

Surgical procedure and antibody administration

The surgical method of producing murine DO was performed as previously described.(12) Briefly the surgical protocol included a longitudinal incision, elevation of the tibia, and blunt separation of the underlying muscles with care not to remove all of the periosteum. The distraction device was attached to the upper part of the tibia, just below the knee, in two parts, with ligature wire (3M Unitek, Monrovia, CA, USA) fixation. A transverse tibial osteotomy was created with the use of a serrated scalpel blade in the proximal diaphysis of the tibia, whereas the fibula was left intact. Both proximal and distal segments were approximated, and alignment of the device was confirmed by X-ray. All animals tolerated the procedure well and were mobile on awakening from anesthesia.

Three hundred fifty microliters of VEGF receptor antibodies was administered intraperitoneally (1.5 mg/ml for both DC101 and MF1 in 1× PBS or DC101 alone) starting at day 3 and were given every third day until death. This injection protocol and dosing was as previously used to obtain receptor inhibition that blocked tumor growth.(16,17) Control animals underwent the same surgery and were injected with mouse nonimmune serum. Active distraction (0.15 mm/d) was initiated on day 7 after surgery. During the active distraction period antibody injections and distractions were done concurrently. Antibody delivery was continued through the end of the consolidation period. Animals that were harvested during active distraction were all killed 1 h after the last activation of the distraction device. No systemic effects on the overall health of the mice were visualized in either the control or the experimental groups.

RNA isolation and ribonuclease protection assay

RNA was isolated by Trizol extraction and quantified using its OD₂₆₀ nm. The integrity of the RNA was visualized by a denaturing RNA gel electrophoresis. To determine the effect of the inhibitors on molecular events of bone formation, the expression of a panel of extracellular matrix protein mRNAs reflective of both chondrogenic and osteogenic cellular activities was initially examined by ribonuclease protection analysis. Because probes for multiple mRNAs are labeled in common and the hybridization of multiple mRNAs is in the same solution, the only potential variable that would effect these measurements would be labeling bias for the individual probe sequences. Thus, the intensity of each of the protected products will be representative of the absolute levels of each of the mRNAs and provides a measure of the absolute amounts of skeletogenic tissue formation. Ribonuclease protection assay (RPA) was carried out as previously described to assess the expression of osteopontin (OPN), bone sialoprotein (BSP), collagen α1 type II (Col 2A1), collagen α1 type X (Col10A1), and osteocalcin (OC). All values were normalized to the expression of the L32 housekeeping gene.(20)

Quantitative real-time RT-PCR

All reagents for the RT-PCR analysis were from Applied Biosystems, and plate assays were read on an ABI 7700 Sequence Detector (Applied Biosystems, Foster City, CA, USA). One microgram of total RNA was used for each preparation of cDNA. All cDNA preparations were generated by random hexamer priming. Reverse transcription reactions were initiated at 25°C for 10 min and subsequently carried out for an additional 60 min at 37°C. The cDNAs were diluted at a ratio of 1:50 in RNase-free water for all PCR reactions. All external and internal primers were from commercial sets available from Applied Biosystemswith the exception of osteocalcin. The amplicon sizes and the 5′ primer of each set are summarized in Table 1. All internal primers were labeled with FAM sequence. Each plate contained two negative controls and a positive control probe. β-actin was used in all experimental samples for normalization of each target sequence. Results of these studies were derived from the average of three replicate experiments. All graphic data for mRNA expression are presented as days after surgery and with numeric values presented as the relative fold expression to the reference (day 0).

The fractional cycle number at which the fluorescence passes the fixed threshold (C_T values) was used for quantification by using a comparative C_T method. This method is described within the Applied Biosystems instructional manual for the instrumentation. In this method, the threshold (C_T values) is determined when fluorescence of probe reaches its exponential phase of accumulation. Sample values are normalized to the threshold value for β-actin (Actb) for each time point: ΔC_T = XC_T (exp) – XC_T (Actb). The C_T value for day 0 was used as a reference. ΔΔC_T = XC_T (exp) – XC_T (exp day 0). The fold change in mRNA expression for each time point was plotted in a graph using day 0 as a reference: 2^{−ΔΔCT(day 0)} = 1 as previously described.(21)

Assessments of bone and vascular formation by μCT

Vasculature formation was assessed by μCT using a Scanco μCT 40 system (Scanco Medical, Basserdorf, Switzerland) at a resolution of 12 μm. The following method, as adapted from Duvall et al.,(22) was used to assess vessel formation. After the animals were killed, the thoracic cavity was opened, and the inferior vena cava was severed. The vasculature was flushed with 0.9% normal saline containing heparin sodium (100 U/ml) through a needle inserted into the left ventricle. The specimens were pressure fixed with 10% neutral buffered formalin. Formalin was flushed from the vessels using heparinized saline, and the vasculature was injected with a radiopaque silicone rubber compound containing lead chromate (Microfil MV-122; Flow Tech, Carver, MA, USA). Samples were stored at 4°C overnight for contrast agent polymerization.

Preceding μCT imaging, the fixators were removed from the mouse, and the hind limbs were dissected, leaving most of the muscles in place over the osteotomy site. The tissue specimens were fixed for 4 days in 10% neutral buffered formalin. Tissues were subsequently treated for 48 h in a formic acid–based solution, Cal Ex II (Fisher Scientific, Pittsburgh, PA, USA), to decalcify the bone and facilitate image thresholding of the hind limb vasculature from the surrounding tissues. The data from these experiments were compared by one-way ANOVA followed by Bonferroni test for the multiple comparisons, using the SPSS11.0 software. p = 0.05 or 0.01 was considered as the significant level.

Separate specimens were used to quantify the mineralized tissue formation within the distraction gap. For these studies, hind limbs were disarticulated immediately after death, and the fixator was carefully removed, leaving the soft tissue covering intact. The specimens were fixed for 3 days at 4°C in neutral buffered 4% paraformaldehyde, after which they were stored in neutral phosphate-buffered saline at 4°C. At the time when the μCT measurements where carried out, specimens were embedded in 2% agarose to provide structural support to the regenerate during μCT scanning. Specimens were scanned at a resolution of 16 μm using a Scanco μCT 40 system (Scanco Medical, Basserdorf, Switzerland). For these measurements, a region of interest (ROI) was defined for each of the experimental sample groups. The ROI was the distance along the diaphyseal axis that spans the region at which the cortex of the native bone (as seen in a transverse section) discontinues proximally to the point at which it continues distally. After determining the ROI for each specimen, the contours of the distraction callus were outlined. A global thresholding algorithm was applied with a fixed, constant threshold for all specimens. This threshold was chosen as the intensity (grayvalue) that corresponded to ∼45% of the average intensity of intact cortical bone in these specimens. Voxels with intensities higher than the threshold were considered to contain mineralized tissue. A constrained 3D Gaussian filter (filter width = 0.8, filter support = 1 voxel) was used to partially suppress image noise. Total bone volume and average mineral density were each compared across groups using a Kruskal-Wallis test (ANOVA by ranks).(23)

Demineralized histology and immunohistochemistry

For histological assessment, the distraction sites with surrounding muscle and soft tissues and were fixed, decalcified, sectioned, and stained as previously described.(24) Serial sections were generated, and slides were taken every 100 μM. Slides were stained with H&E. Similar immunohistological procedures to those that we used previously were carried in this study.(11) Antigens were retrieved by incubating sections at 65°C in 10 mg/ml hyaluronidase for 1 h. Endogenous peroxidase activity was extinguished by incubating twice in 3% hydrogen peroxide for 15 min. Three percent weight of reconstituted nonimmune serum from the animal source in which the secondary antibody was produced was resuspended in PBS and used as a blocking reagent for nonspecific background binding. Sections were incubated with the primary antibody, Sox9:goat polyclonal IgG (1:10,000 1% nonimmune serum PBS solution) for overnight at 4°C. Secondary labeling was accomplished by incubation with the appropriate biotin conjugated secondary antibody (10 μg/ml 1% nonimmune serum PBS solution) at room temperature followed by incubation with Vectastain ABC reagent (Vector Laboratories, CA, USA) for 30 min. Color was developed with the diaminobenzidine (DAB) chromogen kit (Vector Laboratories, CA, USA) for 3 min, checking for color development under the microscope. For each staining run, negative control sections were incubated with 1% nonimmune serum PBS solution without the primary antibody.

VEGF receptor and ligand expression during DO

The expression of all four VEGF mRNAs was induced throughout the period of distraction and consolidation (Fig. 1). The induction of VEGFD preceded that of all the other ligands showing a definitive ∼6-fold peak in expression at the end of the distraction period (day 17), after which its expression decreased. In contrast, the expression of the other ligands showed only ∼2-fold induction that peaked after the consolidation period had been initiated at day 20. Whereas VEGFA showed the same pattern of induction as VEGFB and VEGFC, its elevated expression did not decrease within the 31-day experimental period.

All three VEGF receptors also showed an induction of their expression over the time course of DO. The expression of both VEGFR2 and VEGFR3 generally paralleled that of seen for the ligands, showing induction beginning at ∼10 days after surgery (3 days after the initiation of active distraction) that continued to rise until they peaked during the consolidation period (day 20), after which time they began to decrease. It is also interesting to note that the VEGR1 receptor reached its peak expression by the end of the distraction period (day 17) and showed a more constant level of induction throughout the period of consolidation in a similar manner to the VEGFA ligand.

Radiographic assessments of skeletal and vasculature tissue formation

To elucidate the functional roles of the VEGF receptors in promoting angiogenesis and osteogenesis, antibody blockade against VEGFR1 and VEGFR2 was used to inhibit both receptors together or VEGFR2 alone. Faxitron radiographic analyses were used as an initial qualitative assessment of skeletal tissue formation. Control specimens showed robust bone formation by the end of consolidation (day 31). In the experimental animals, however, blockade of VEGFR2 alone or both VEGFR1 and VEGFR2, respectively, showed a progression in the inhibition in of bone formation within the osteotomy gap (Fig. 2A).

μCT analysis was next carried out to provide both a structural and quantitative assessment of mineralized skeletal tissue formation during DO. Antibody blockade of both VEGFR2 and both VEGFR1and 2 led to the selective loss of external periosteal bone formation. In contrast, there was a more progressive loss of trabecular bone formation in the intramedullary space that was seen between the single and double antibody treatments (Fig. 2B). Quantitative measurements made from the μCT assessments showed ∼2- and ∼6-fold less bone formation in comparison with controls with the single and double antibody treatments, respectively (Fig. 2C). The relatively small sample size used in this analysis, however, did not show significant quantitative differences in the amount of mineralized tissue accumulated in the gap between the single and double antibody treatment, although both were statistically less than the control.

μCT imaging after vascular Microfil perfusion was used to assess the effect of VEGF receptor blockade on the formation of new blood vessels in the distraction gap (Fig. 3). A similar graded effect between the single and double antibody blockade treatments was observed for vascular formation. Decreasing values for vessel volume, number, volume/total volume, connectivity, density, and surface area, with an increase in vessel separation, was seen in the VEGFR 1 and 2 antibody–treated mice compared with the VEGFR2 and nontreatment controls. In this analysis, even with the small sample size, all these indices showed significance both between them and with the controls (Fig. 3B; Table 2).

Molecular assessments of skeletal cell development

The molecular mechanisms that were regulating the cellular development within the regenerate tissues were assessed by two different approaches. RPA was used to measure both the temporal progression and quantitative amounts of cartilage and bone formation. These results, which are seen in Fig. 4A, confirmed the radiological analysis that the lack of skeletal tissue formation in the experimental groups was reflected in the diminished cellular activities of skeletogenic cells. These results further showed that, whereas the expression of the osteogenic mRNAs was diminished, the chondrogenic mRNAs showed an earlier and more sustained expression in the single antibody–treated animals, although at a much lower level than osteogenic expression.

To quantify how the inhibition of VEGFR signaling and the lack of angiogenesis affected the temporal progression of skeletogenic cell differentiation, real-time qRT-PCR was used to measure the relative levels of BMP-2, Col2A1, and osteocalcin mRNA expression as representative markers for differentiation of chondrogenic and osteogenic cells (Fig. 4B). BMP-2 expression was analyzed as a means of determining the degree to which inhibition of angiogenic development affected the morphogenetic signals that drive skeletal tissue formation. This analysis showed that BMP-2 levels peaked at ∼6-fold the levels of unorperated bones during the period of active distraction. The levels slightly decreased but remained elevated ∼4-fold throughout the remainder of the experimental period. Whereas both antibody treatments led to decreased levels in the expression of BMP-2, its level of expression remained ∼3-fold above that seen in the unoperated tissues. Interestingly, the double antibody treatment did not show lesser amounts than the single treatment, and a sharp increase in the levels of BMP-2 expression was seen in the double antibody–treated tissues at the end of the experimental period.

The patterns of chondrogenic and osteogenic mRNA expression seen for the qRT-PCR findings showed that the relative levels of expression of Col2A1 for the single receptor blockade was greatly increased during the active distraction period, after which it showed a rapid decrease during consolidation. In contrast, the double blockade treatment resulted in a small transient elevation at the very beginning of distraction but thereafter showed a steady decline over the remainder of the time course. Interestingly, the control samples showed an induction of Col2A1 expression throughout the consolidation period consistent with both our previous results(12) and the RPA analysis shown in Fig. 4A.

Osteocalcin expression, on the other hand, had a very sharp and almost 40-fold increase in its expression in the period of active distraction. Thereafter, the levels rapidly declined, although it continued to show a ∼5-fold elevation throughout the consolidation period. In an inverse manner to the expression of Col2A1, the single blockade samples showed elevated osteocalcin expression during the consolidation period, whereas the double antibody blockade–treated samples had no substantial increase at any time throughout DO (Fig. 4B).

The effect of the antibody blockade on the recruitment and differentiation of skeletogenic stem/progenitor cells was next assessed by examining a set of transcription factors that are known to be determinants for the chondrogenic and osteogenic lineages (Fig. 5). The four transcription factors that were compared included the following: SOX 9, which commits skeletal stem/progenitor cells to the skeletal lineage and is common to both chondrogenic and osteogenic cells; SOX 5, which subsequently contributes to the commitment of cells to the chondrogenic lineage; osterix, which helps commit the lineage to osteogenesis; and RUNX2, which is involved in both terminal hypertrophic chondrocyte differentiation and is necessary with osterix to promote osteogenesis.(25) The results for the expression of osterix and Runx2 seen in the top two panels of this figure are parallel to those seen in the terminal osteocalcin marker for the differentiation of osteogenic cells. There is almost a complete lack of induction of osterix expression in the double antibody–treated samples. These results would indicate that the more complete blockade of angiogenesis produced by the double antibody treatment blocks the progression of any of the recruited cells to an osteoblastic lineage. The expression of SOX9 and SOX5 expression are depicted in the bottom panels. The very high levels of expression of both SOX9 and SOX5 mRNAs in both the single- and double antibody–treated samples at the end of the latency period suggests that skeletal stem/progenitor cells can still be recruited but they remain as common progenitors or selectively progress toward a chondrogenic phenotype when vascularization is compromised.

Histological assessments of skeletal cell development

The last set of studies presents a histological and immunohistological analysis of the specimens as a means of validating the cellular activities that were extrapolated from the various mRNA assessments. Micrographs of control regenerate samples taken from the intramedullary space of both 21- and 31-day postsurgery specimens are first presented to highlight the various biological processes and the morphogenesis of the regenerate tissues that are formed (Fig. 6). The regenerate tissue in the control specimens is characterized by robust trabecular bone formation within the intramedullary space. As the tissue is progressively lengthened, new bone formation proceeds from the proximal side (right side of the micrograph) inward to the distal end of the gap (left side of the micrograph). This can be seen by comparing the micrographs from the two different postsurgical times. As observed in this comparison, the formation of mature trabeculae arises from condensation at the proximal side, and as the tissue develops distally, dense layers of cells became aligned in columns, some of which have central cores of cartilage cells. The foci of chondrocyte within the cores of the condensing tissues are seen in the higher-magnification micrographs (Fig. 6B). In between these columns of new bone tissue, new blood vessels are formed that are lined with endothelia-appearing cells with well-defined concave-shaped red blood cells (RBCs) contained with the luminal spaces of these large vessels (Fig. 6C). The aligned nature of the newly formed trabeculae to the long axis of the bone can be further shown by assessing the underlying collagen fiber orientations with polarized light (Fig. 6D).

The global effect of VEGFR blockade on tissue formation is presented in Fig. 7A. As can be seen in these micrographs, considerable amounts of tissue formation develop external to intramedullary space of the control specimens. The tissues that formed externally to the distraction gap form a shell of bone that bridges the gap with small areas of cartilage in the central regions within the external callus regions. By 31 days after surgery, no cartilage tissue was seen in the medullary space of the control specimens in contrast to the control specimens at 21 days. In comparison, the single- and double antibody–treated samples developed much lesser amounts of intramedullary trabeculae and external callus tissues. The most severe diminishment in tissue regeneration was seen in the midregions of the double antibody–treated tissues. Both the single and double antibody treatment specimens also showed only a thin edge of developing bone along the surface of the external callus proximal and distal to the gap but not bridging the distraction gap at all. Both single and double antibody blockade treatments developed areas of cartilage tissue both within the intramedullary space, with the single treatment showing lesser amounts than the double treatments. The double antibody blockade treatment, however, showed extensive areas of hypertrophic-looking chondrocyte, whereas the proximal regions of the gap was mainly fibrous and fat tissues (Fig. 7B).

It is interesting to note that, whereas tissue from both the single- and double antibody–treated animals showed bands of aligned fibrous tissues surrounding external areas of the gap, no well-developed woven bone formation was histologically seen in the regenerate that formed in osteotomy gaps. Although lesser amounts of mineralization of these tissues must have taken place based on the comparison with the μCT data seen in Fig. 3, no tissues with a woven bone or trabecular appearance were seen in the histological analysis of these samples. The histological results suggest that the mineral deposition is most likely associated with areas of hypertrophic cartilage tissue development.

Immunohistological analysis was next carried out for SOX9. Micrographs depicting the H&E staining and matched serial sections developed with anti-SOX 9 antibody are seen in B and C. These analyses were carried out to assess the spatial distribution and histological nature of cells that commit to the skeletal cell lineages in control DO tissues and in the regenerate tissues in which VEGFR activity has been inhibited. In the regenerates tissues in the distraction gaps from the control animals, antibody staining was predominantly localized within the nuclei of large round chondrocyte-looking cells that were spatially either aligned in long columns in the intramedullary gap regions or in the external callus regions of the in the middle regions of the gap. The most intense staining was detected in the cells from the single- and double antibody–treated animals. It is interesting to note that their were many more labeled cells in the control specimens from 21 days and much fewer and much less intensely reacting cells were observed in the tissues from the controls at 31 days.