Bone vascularization and trabecular bone formation are mediated by PKBalpha/Akt1 in a gene dosage dependent manner: In vivo and ex vivo MRI

Abstract

PKBalpha/Akt1, a protein kinase, is a major mediator of angiogenic signaling. The purpose of this study was to determine the role of PKBalpha/Akt1 in bone vascularization and development. For that aim, macromolecular dynamic contrast enhanced MRI (DCE-MRI) was applied to examine in vivo vascular changes in long bones of 40-day-old growing PKBalpha/Akt1 deficient, heterozygous, and wild type mice. Ex vivo μMRI and μCT were applied to monitor the impact of PKBalpha/Akt1 gene dosage on trabecular bone formation during endochondral bone growth. PKBalpha/Akt1 deficient mice and remarkably also heterozygous mice, showed significantly reduced blood volume fraction in the humerus compared to wild type mice. Moreover, PKBalpha/Akt1 deficient mice showed a more severe vascular deficiency with reduced permeability. μCT and μMRI of trabeculae revealed impaired bone formation in both PKBalpha/Akt1 deficient and heterozygous mice, whereas cortical bone parameters were only reduced in PKBalpha/Akt1 deficient mice. Reduction of metaphyseal blood vessel invasion, concomitant with aberrant trabeculae and shorter long bones, demonstrates a gene dose dependent role for PKBalpha/Akt1 in regulation of overall size and endochondral bone growth. MRI proved to provide high sensitivity for in vivo detection of subtle gene dose effects leading to impaired bone vascularity and for uncovering changes in trabecular bone.

Introduction

Endochondral bone formation, the replacement of avascular cartilage by vascularized bone, is essential for longitudinal bone growth during vertebrate development. During this process, a cartilaginous plate (growth plate) is generated between the shaft of long bones (diaphysis) and their ends (epiphysis). At the cartilaginous epiphyseal plate, sequentially chondrocyte proliferation, hypertrophy, apoptosis and invasion of vasculature occurs, forming primary trabecular bone (1-3). Newly formed blood vessels, invade the region between hypertrophic chondrocytes and the newly formed bone matrix at the base of the metaphysis, and provide nutrients for the highly specialized cells, involved in the regulation of bone formation. Angiogenic growth factors such as vascular endothelial growth factor (VEGF), were shown to affect bone development by triggering blood vessel invasion (3-9). Suppression of VEGF-driven angiogenesis during endochondral bone formation has been shown to impair trabecular bone formation (4).

PKBalpha/Akt1, an intracellular protein kinase, acts downstream of VEGF stimulation of the VEGF receptor in endothelial cells, through phosphatidyl-inositol 3-kinase (PI3K) signaling. PKBalpha/Akt1 is considered to be a major mediator of signaling of angiogenic growth factors, affecting endothelial cell survival, proliferation and differentiation (4,10). In addition to angiogenesis, PKBalpha/Akt1 regulates many other cellular and physiological processes, such as glucose metabolism, transcription, cell cycle regulation, survival and inflammation. PKBalpha/Akt1 deficient mice are smaller, with increased neonatal mortality along with disordered fetal vasculature and placental hypotrophy (11,12). Moreover, these mice were reported to exhibit bone mineralization defects characterized by decreased length and bone mass of long bones (13,14).

Since postnatal longitudinal bone growth requires infiltration and expansion of the newly formed blood vessels and on the other hand, PKBalpha/Akt1 mediates intracellular signaling of angiogenesis, we postulated that a vascular deficiency at the site of the long bones could contribute indirectly to impaired bone development in PKBalpha/Akt1 deficient mice. This led us to study postnatally, the vascularization and development of long bones in these mice during endochondral bone growth. In particular, in view of the remarkable dose-dependent sensitivity to VEGF signaling during development (heterozygous in vivo mortality of VEGF deficient; (15)), we applied here MRI as a sensitive and quantitative tool for comparing the impact of homozygous and heterozygous deficiency of PKBalpha/Akt1.

In this work, we used dynamic contrast enhanced (DCE) MRI with macromolecular contrast media for quantitative, noninvasive, functional analysis of the microcirculation within the long bones in general, and neovascularization of the metaphysis, the growth zone, of growing long bones, in particular. In the past, changes in new vessel formation at the growth plate, during endochondral bone formation, have been evaluated using post mortem immunohistochemistry (4), evaluating proliferation, apoptosis and microvessel density. In contrast with immunohistochemistry, macromolecular DCE-MRI provides information on vascular functionality and allows in vivo follow up. The use of dual-modality contrast material enabled histological validation of the DCE-MRI data.

Previously, we reported noninvasive MRI, as well as fluorescence microscopy validation of vascular development and associated hyperpermeability in implantation (16), tumors (17), ischemic injury (18), and ovarian xenografts (19) using Gd-DTPA bound to bovine serum albumin and biotin (biotin-BSA-GdDTPA). Recently biotin-BSA-GdDTPA was applied for analysis of bone vascularization during tumor progress (20). Here, we show that biotin-BSA-GdDTPA, because of its selective extravasation from permeable vessels and its slow diffusion and clearance from the extracellular space, allowed high-resolution detection of disturbances of metaphyseal blood flow in PKBalpha/Akt1 deficient mice. This indicates that macromolecular DCE-MRI could be used as an early indicator of impaired vascular function in diseases where altered angiogenesis causes impaired skeletal growth.

Thus, the purpose of this study was to apply MRI as a sensitive, quantitative tools to determine vascular function in the long bones of growing PKBalpha/Akt1 deficient (−/−) and heterozygous (+/−), mice and to study the impact of PKBalpha/Akt1 gene dosage on trabecular bone formation during endochondral bone growth.

Methods

Animals

All animal experiments were approved by the Weizmann Institutional Animal Care and Use Committee. Male PKBalpha/Akt1 wild type (+/+), heterozygote (+/−), or knockout (−/−) mice were studied by MRI on postnatal day 40. At 40 days of age, PKBalpha/Akt1 −/− were smaller, whereas +/− are intermediate compared to the +/+ animal size (data not shown). This age was selected for in vivo imaging because animals are still growing, their growth plate is open and they are more easily accessible for intravenous (iv) contrast injection than younger animals.

Contrast material

Biotin-BSA-GdDTPA (about 82kDa) was prepared as reported previously (17) and injected through a tail vein catheter as bolus (0.5mg/g).

In vivo DCE-MRI studies

The transparency of bones to MRI enabled us to evaluate in vivo the vasculature within the developing long bones. MRI experiments (n=7 for each group) were performed at 9.4T on a horizontal Biospec (Bruker, Germany) using a linear resonator for excitation and an actively decoupled 2cm surface coil for detection. Mice were anesthetized (intraperitoneal; 75mg/kg ketamine, Fort Dodge Laboratories, IA, USA; and 3mg/kg xylazine 2%, VMD, Belgium) followed by subcutaneous addition of about 30% of the initial dose. Anesthetized mice were positioned, so that left front leg, and particularly the humerus, was at the center of the surface coil. The tail vein was catheterized with homebuilt catheters fitted with heparin washed needle, for administration of the contrast media. Body temperature of the animals was controlled using a warming water blanket (Bruker, Germany).

At the end of the MRI experiment, 30min after contrast injection, bovine serum albumin (BSA) labeled with rhodamine (BSA-ROX; 1.4μmol/kg; Molecular Probes, OR, USA), as an early vascular marker, was iv injected via a tail vein catheter 3-5min prior to animal sacrifice, as reported previously (21).

DCE-MRI data acquisition

3D gradient echo (3D-GE) images of the left front limb were acquired before and sequentially for 30min after iv injection of biotin-BSA-GdDTPA. The 3D field of view (FOV) of the image covered the proximal front limb. A series of variable flip angle precontrast T₁-weighted 3D-GE images were acquired to determine the endogenous precontrast R₁. Imaging parameters: precontrast flip angles 5°, 15°, 30°, 50°, 70°; postcontrast flip angle 15°; TR 10ms; TE 4ms; two averages; spectral width 50,000Hz; matrix 128×128×64; zero-filled to 256×256×128; FOV 30×30×15mm; resolution isotropic 234μm; acquisition time 163s.

DCE-MRI data analysis

Pixel-by-pixel analysis was done on a pc using MATLAB software (MathWorks Inc., MA, USA) to generate concentration maps of biotin-BSA-GdDTPA for selected slices containing the humerus of the 3D datasets as described before (17,22). First, precontrast longitudinal relaxation rate (R₁ pre) maps were derived from the variable flip angle data by nonlinear best fit to Eq.[1]:

$$I=\frac{M_0\sin \alpha (1-e^{-\mathit{TR}•R_1\text{pre}})}{1-\cos \alpha •e^{-\mathit{TR}•R_1\text{pre}}}$$ [1]

Where I is the signal intensity as a function of pulse flip angle α, TR is the repetition time (10 ms), and the preexponent term M₀ includes the spin density and the T₂ relaxation, which are assumed to be constant. Then, post contrast R₁ values (R₁ post) were calculated from the precontrast and postcontrast signal intensities ( I_pre and I_post, respectively, acquired with a flip angle of 15°). Finally, concentration maps were calculated based on the relaxivity (R) of biotin-BSA-GdDTPA (99 mM⁻¹s⁻¹ per BSA, at 9.4T) (17,22).

During the first 15min after administration of the contrast media, its concentration in circulation was stable (21) and contrast accumulation is linear in regions of permeable vasculature but not yet noticeably affected by interstitial convection (22). Therefore, linear regression of the dynamic change in concentration during the first 15min postcontrast was used for the derivation of two vascular parameters, for selected ROI, the humerus and for each pixel in parametric maps (21) (Fig. 3b). The blood volume fraction (fBV), was derived from the ratio between the extrapolated concentration of biotin-BSA-GdDTPA at the time of administration and the concentration in the blood calculated from mean concentration values at selected ROIs within the cephalic vein. The permeability surface area product (PS; min⁻¹) is derived from the initial rate of accumulation of the contrast material, at a selected ROI containing the left humerus enhancing between 0 and 15min, normalized to the initial blood concentration. PS reflects the extravasation of macromolecules such as albumin from blood vessels and their accumulation in the tissues.

Figure 3.

Mapping vascular parameters in PKBalpha/Akt1 mice by DCE-MRI. Single sections derived from 3D gradient echo data containing the humerus, MIP, fBV maps and PS maps. Linear fit of the first 15 minutes (trend lines, see figure 1) was used for derivation of blood volume fraction and permeability surface area product (fBV= tissue concentration at time zero divided by blood concentration; [C]= tissue concentration of contrast agent at time zero; [Cblood]= blood concentration of contrast agent at time zero; PS= the slope of the normalized contrast concentration).

Numeric values for fBV and PS were calculated by selecting ROIs that delineate the humerus. For presentation of parametric maps (Fig. 3), the stacks of fBV maps (generated for slices of interest) were projected to show the mean value in each pixel in the axial plane.

Histology, immunohistochemistry and fluorescence microscopy

For histological validation of the DCE-MRI, dual labeling was performed using BSA-ROX and biotin-BSA-GdDTPA. The right humeri of 40-day-old mice were fixed (Carnoy’s solution for fluorescence), decalcified and embedded in paraffin, sectioned serially at 4μm thickness and stained for the biotinylated contrast agent with avidin-FITC (Molecular probes, CA, USA) as previously described (21). BSA-ROX injected just before collection of the tissue, remained intact throughout the processing mentioned above. For immunohistochemistry, tissue samples were fixed in 4% PFA, decalcified, embedded. Subsequently, tissue sections were stained with anti CD34 (Cedarlane laboratories, Canada). For H&E staining (histomorphometric analysis), the right humeri (n=4 for each group) were fixed in 4% PFA, decalcified, embedded and stained with hematoxilin and eosin (H&E).

Morphometric analysis of histological sections

H&E stained histological sections of bone were analyzed by basic stereological approach (23). For each group (n=4), decalcified H&E sections of the humerus (2-3 sections per animal) were imaged with Eclipse E800 microscope and analyzed by computer-assisted image analysis using Image Pro Plus 5.0 Software (Media Cybernetics; Eclipse E800 microscope connected to Nikon Digital Camera DXM 1200). The proximal humerus was studied not more than 500μm distal to the growth plate (distal to the zone of calcification). A grid was placed on square fields of < 0.25mm², for each section and each parameter producing a particular point count. For quantitative image analysis, three parameters were measured: (i) the percentage of trabeculae, which represents the number points on the grid corresponding to trabeculae, divided to the total number of points (trabeculae representing bone matrix, calcified cartilage matrix with osteoblasts and osteoclasts) ; (ii) the percentage blood vessels (sinusoids; delimited by endothelial cells containing red blood cells) and (iii) the percentage of bone marrow (interspersed between trabeculae and blood vessels containing cells of blood cell synthesis) were calculated the same way.

Bone morphometric analysis by μCT and μMRI analysis

μCT

Left humeri of (n=4 for each group) were dissected, cleaned and stored at -20°C until tested. Bones were thawed at room temperature, and scanned by CT (eXplore Locus SP, GE Healthcare, Canada) with 80kV x-ray voltage, 80μA current and 400ms integration time. Three-dimensional 8μm isotropic images were obtained using μCT Images were reconstructed and thresholded to distinguish bone voxels with MicroView software version 5.2.2 (GE Healthcare, Canada). Analysis of cortical bone (bone mineral density, bone mineral concentration and mean thickness) was performed on a transverse section of the diaphysis equivalent to 3.5% of the whole bone length, starting immediately adjacent and distally to the deltoid tuberosity. Analysis of trabecular bone was performed on a transverse section of the metaphysis equivalent to 3% of the whole bone length, starting immediately adjacent and distally to the growth plate. ROI consisted of the proximal metaphyseal region and used for measurement of bone volume fraction (BVF), number of trabeculae and their spacing.

μMRI

Trabecular bone was also measured by μMRI of the femur. Femora of the same mice as for μCT, were used and for each group one extra animal was added (n=5 for each group). The bones were placed in 5mm NMR tubes with 1.5mM Gd in saline, placed under vacuum to eliminate air bubbles and imaged at 9.4T with FLASH 3D on a Vertical bore Avance III DRX 400MHz NMR spectrometer (Bruker, Germany), equipped with microimaging probe and a 5mm insert coil. Images of distal femora of these mice were acquired with following imaging parameters: flip angle 15°; TR 15ms; TE 3.3ms; 4averages; spectral width 100,000Hz; matrix 512×256×512; FOV 10×5×10mm; 20μm isotropic resolution.

Data were analyzed by the Virtual Bone Biopsy (VBB) processing algorithm (24). The 3D datasets were first segmented by hand to obtain the trabecular bone region of interest. A custom user interface was used for this purpose, which allowed the user to manually outline the trabecular bone region (using polygons) in a few axial reference slices. The program then linearly interpolated between the reference slices to achieve a 3D segmentation. The segmented volume was then subjected to the Virtual Bone Biopsy (VBB) processing algorithm, comprising the following steps: (1) correction for coil shading; (2) inversion, interpolation, and binarization; (3) 3D skeletonization; and (4) topological classification via digital topological analysis (DTA). The DTA step results in a voxel-by-voxel classification of the 3D skeleton into plates, rods, and junctions of various types. From these, composite structural parameters surface-to-curve ratio (S/C) and erosion index (EI)) were obtained. The S/C is an estimation of the number of plate-like voxels divided by the number or rod-like voxels in the skeleton. It is generally understood that the trabecular network’s strength is impaired when plates are more eroded and converted into rods. Therefore, a higher S/C value suggests stronger bone. EI is a similar parameter, also a ratio of the sum of parameters expected to increase with reduced trabecular network’s strength divided by the sum of those expected to decrease. Therefore lower EI suggests stronger bone. Since the distal femora varied widely, in terms of length and shape, the use of objective criteria for selecting the portion of the bone to be analyzed was required. While the cross-sectional segmentation was straightforward to perform manually, the choice of longitudinal region of interest was less clear. Therefore, in each sample, the thirty contiguous slices with highest BVF were automatically selected for analysis, and the parameters were computed only over these slices.

Statistical Analysis

One-way ANOVA was used for the analysis of significance of different parameters of the three groups, PKBalpha/Akt1 +/+, +/− and −/− mice. When the differences were found statistically significant (P< 0.05), post hoc test was performed using Tukey HSD test to control for familywise error rate (FEWR). The mean value ± SD is described for each group.

The authors had full access to the data and take responsibility for its integrity. All authors have read and agreed to the manuscript as written.

Results

PKBalpha/Akt1 controls bone vascularization in a gene-dosage dependent manner

Blood volume fraction (fBV) and permeability (PS) in the humerus and its proximal growth plate were measured in vivo by DCE-MRI using pharmacokinetic analysis of macromolecular biotin-BSA-GdDTPA. Two regions of interest (ROI) were analyzed (Fig. 1, 2) and pixel by pixel analysis was done (Fig. 3) to derive values and maps of blood volume (fBV) and vessel permeability (PS). The first ROI, the humerus (Fig. 1a,b), a highly vascularized long bone, showed a high initial contrast enhancement, reflected by a high fBV in the wild type (0.0785±0.013). The fBV was significantly reduced in PKBalpha/Akt1 deficient (−/−) and heterozygous (+/−) mice compared to the wild type mice (0.0207±0.014 and 0.0392±0.012 respectively; Fig. 2a). This result indicates gene dosage dependent effects with reduced (micro)vascular density, not only within the humerus of PKBalpha/Akt1 −/−, but remarkably also in PKBalpha/Akt1 +/− mice.

Figure 1.

In vivo dynamic contrast-enhanced MRI (DCE-MRI) of the role of PKBalpha/Akt1 in bone vascularization. (a) Consecutive MRI images of the front limb of growing male mice, were acquired before and after administration of macromolecular contrast material (biotin-BSA-GdDTPA; intravenous, via a tail vein catheter). MRI slice showing the humerus (H) and the growth plate area (G) as the regions of interest (ROIs) for analysis. (b) DCE-MRI for 30 minutes after contrast injection (10 time points) of the humerus normalized with contrast concentration in the cephalic vein at each time point. (c) Dynamic 3D imaging included follow-up of contrast enhancement for 30 minutes post contrast (10 time points) for the area of the growth plate. For both (b) and (c), fractional blood volume (fBV; intercept with time zero) and permeability surface area product (PS; the slope) values were calculated from a linear regression of the first 15min for each group. (−/− = PKBalpha/Akt1 knockout; +/− = heterozygote; +/+ = wild type; n=7 for each group).

Figure 2.

Quantification of vascular MRI parameters in PKBalpha/Akt1 bones calculated from a linear regression of the first 15min in DCE-MRI. (a, c) blood volume fraction (fBV) and (b, d) permeability surface area product (PS) of the humerus (a, b) and the region of the proximal growth plate of the humerus (c, d). (a,c) fBV: the ratio between the extrapolated concentration of biotin-BSA-GdDTPA at the time of administration and the concentration in the blood. This parameter indicates the fraction of blood volume in the tissue. (b,d) PS: the slope of this linear regression (min⁻¹), measuring the extravasation of macromolecules out of blood vessels. (mean ± SD; * P<0.05; −/− = PKBalpha/Akt1 knockout; +/− = heterozygote; +/+ = wild type; n=7 for each group).

The slope of the linear regression of the first 15min, the PS (Fig. 1a), showed contrast accumulation over time due to high vascular permeability of sinusoids in the bone in wild type mice (0.0049±0.0009). The PS was significantly reduced in PKBalpha/Akt1 −/− (0.0017±0.0007). No significant difference in PS was found in +/−, relative to wild type mice (Fig. 2b). The concentration of contrast agent in the blood remained constant over the first 15min and no difference in dynamics or concentration was found between the groups (data not shown). The slope of the contrast behaviour curve of the later data points after 15min, was only significantly different for PKBalpha/Akt1 +/− mice. Though each separate timepoint of PKBalpha/Akt1 −/− mice after 15min was significantly different from PKBalpha/Akt1 +/+ mice. Heterozygous PKBalpha/Akt1 +/− mice showed significantly different contrast changes for the two last time points and showed intermediate behavior at earlier time points (16-22 min).

The second ROI analyzed, the proximal growth plate region of the humerus (Fig. 1c), showed lower initial contrast enhancement and thus appeared to be less vascularized than the whole bone (fBV; 0.0184±0.0050). The lower fBV values are consistent with inclusion of the avascular epiphysial plate in this ROI. The blood volume fraction, fBV, was significantly decreased in PKBalpha/Akt1 −/− and +/− mice (0.0014±0.0008 and 0.0037±0.0021 respectively) compared to the wild type mice, indicating reduced (micro)vascular density in the area of the epiphyseal plate of −/− and +/− mice (Fig. 2c). Vascular permeability (PS) derived from the first 15min, showed the same contrast accumulation over time as over the whole humerus, although the value for the wild types (0.0012±0.0003) was lower than for the entire humerus. The PS was significantly reduced in PKBalpha/Akt1 −/− mice (0.0003±0.0001), but no significant difference in permeability was found in +/− mice (Fig. 2d), relative to wild type mice. The linear regression of the first 15min postcontrast was used for calculation of fBV (Fig. 1b) on a pixel-by-pixel basis and generation of fBV and PS maps, showing a reduced (micro)vascular density in both PKBalpha/Akt1 −/− and +/− humeri (Fig. 3).

Biodistribution of macromolecular contrast agent, immunohistochemistry and microscopic validation by histomorphometric analysis, of the proximal humeral metaphysis of PKBalpha/Akt1 40-day-old mice

Histological staining of biotin-BSA-GdDTPA, was used for validation of the microscopic biodistribution of the contrast material within the humerus. Elevated accumulation of biotin-BSA-GdDTPA was demonstrated beneath the epiphysial plate (G; Fig. 4), corresponding to the high permeability of the sinusoid blood vessels, in the bone of wild type mice, as revealed with DCE-MRI. Large blood vessels could be observed by BSA-ROX staining, as an early vascular marker (Fig. 4b, c). Reduced vessel density, was seen in PKBalpha/Akt1 −/− as well as +/− (Fig. 4c), matching with a reduced (micro)vascular density in these humeri detected in vivo by MRI (Fig. 1, 2). In addition, microvascularity was evaluated using the CD34 monoclonal antibody, pointing out newly formed blood vessels and capillaries distally to the epiphyseal plate (Fig. 4d). Using similar exposure time, no fluorescence could be detected in sections from PKBalpha/Akt1 −/− mice, whereas sections from PKBalpha/Akt1 +/− mice showed intermediate signal.

Figure 4.

Microscopic histological analysis of the distribution of the macromolecular contrast agent. (a) (Left) Surface projection μCT of wild type proximal humerus (see also figure 7) indicating the growth plate (G) separating the proximal epiphysis and metaphysis of the humerus. (b) Composite image of 40x magnification of wild type proximal humerus indicating the region shown in (c) containing the epiphysial plate (G). Tissues were retrieved 30 min after injection of biotin-BSA-Gd-DTPA (green; stained for biotin using avidin-FITC), and immediately after injection of BSA-ROX (red). (c) Blood vessels only are stained with BSA-ROX (red; top row), whereas biotin-BSA-GdDTPA (green; central row) partly leaked out of the sinusoids close to the growth plate (G). (d) CD34 staining for newly formed blood vessels and capillaries. PKBalpha/Akt1 deficient metaphysis (−/−) shows no staining of CD34 in an exposure time that gives in wild type (+/+) staining distal to the growth plate (G). Heterozygous (+/−) staining is intermediate between +/+ and −/− (arrow in the −/− panel indicates an area where low fluorescence could be detected at a higher exposure time, which over saturated the fluorescent signal of the +/+ section).

Histomorphometric analysis of the left proximal humeral metaphysis (Fig. 5) revealed lower trabecular bone and lower blood vessels density along with elevated bone marrow content in both PKBalpha/Akt1 −/− and +/− relative to +/+ mice, while no significant difference was found between PKBalpha/Akt1 +/− and −/− (Fig. 6). These results are consistent with the reduced (micro)vascular density observed by DCE-MRI (Fig. 1-3). In addition, humeral length of 40-day-old PKBalpha/Akt1 −/− was significantly reduced compared to +/+ length, whereas +/− humeral length was significantly different from both −/− and +/− and intermediate between both (data not shown).

Figure 5.

Setup of μCT and μMRI analysis of the bone architecture. (a) For histology, humeri were cleaned, fixed, decalcified, embedded in paraffin, sectioned and stained with H&E. Histomorphometric analysis was performed on H&E slides of the left proximal metaphysis (region just beneath the growth plate) of the left humerus (see Fig. 6). (b) For μCT, right humeri were cleaned of soft tissue and stored at -20°C. After thawing at room temperature, bones were placed with 4 bones in a sample holder, immersed with physiological saline and immobilized with foam rubber, and scanned. Right proximal metaphyseal and cortical area of the humeri were used for trabecular and cortical bone analysis respectively (see Fig. 7 and Table 1). (c) As alternative analysis of trabecular bone, right femora were taken for μMRI. They were cleaned, placed with 4 bones in 5mm NMR tubes separated by a fabricated holder and immersed with 1.5mM GdDTPA in saline and placed under vacuum to eliminate air bubbles. The bones were imaged at vertical 9.4T magnet and the distal metaphyseal trabeculae were characterized using a Virtual Bone Biopsy processing algorithm (see Fig. 7 and Table 1).

Figure 6.

Histological analysis of hypovascularity in humeri of male 40-day-old PKBalpha/Akt1 mice. (a) Histomorphometric analysis on (b) H&E histological slices of the left proximal metaphysis of the humerus (mean ± SD; * P<0.05; −/− = PKBalpha/Akt1 knockout; +/− = heterozygote; +/+ = wild type; n=4 animals for each group, 2-3 sections per animal; G=growth plate; M=metaphysis).

PKBalpha/Akt1 controls trabecular bone sedimentation and structure in a gene-dosage dependent manner

Since sedimentation of minerals, and thus formation of new bone, occurs at the growth plate through metaphyseal blood vessels, giving rise to trabeculae, we focused on trabecular bone analysis. μCT images of left humeri (Fig. 5b) showed impaired trabecular bone parameters at the proximal metaphysis at the diaphysis in PKBalpha/Akt1 −/− and +/− mice. Cortical bone parameters were found only significantly different in PKBalpha/Akt1 −/− mice. Trabecular bone analysis revealed a lower bone volume fraction (BVF), a lower trabecular number and a higher trabecular separation in both PKBalpha/Akt1 −/− and +/− mice. In addition, cortical bone showed decreased bone mineral density, mid-diaphyseal cortical thickness and bone mineral density in - /-, but no reduction was demonstrated in +/− (Fig. 7; Table 1).

Figure 7.

Ex vivo bone analysis of 40-day-old male mice (a) μCT analysis of proximal humerus. Upper panel, surface projection; metaphysis (upper arrow); diaphysis (lower arrow). Middle panel, MIP of metaphyseal trabeculae of the humerus (ROI for analysis). Lower panel, cortical bone at mid-diaphysis showing the nutrient foramina. (b) Distal femora of the same animals imaged by μMRI. Arrows on the longitudinal view (lower) show the region of the epiphysis and metaphysis; slices of MR images of the epiphysis (upper) and metaphysis (middle) illustrate the trabecular bone. The region analyzed by Virtual Bone Biopsy (VBB) is shown between two lines (see Table 1). (c) Surface renderings of the metaphyseal region viewed from the top. Skeletonization converts plates to surfaces (white, surface edges red) and rods to curves (blue). μMRI of trabecular bone of the femoral metaphysis was analyzed using the VBB (see Table 1).

Table 1.

Ex vivo μCT and μMRI analysis of PKBalpha/Akt1 long bones

	−/−	+/−	+/+
μCT Humerus (n=4 for each group)
Trabecular bone analysis
Bone volume fraction (BVF)	0.243± 0.018*	0.256± 0.023*	0.410± 0.007
Trabecular number	7.8± 1.5*	7.4± 0.6*	14.6± 1.7
Trabecular separation (μm)	117± 9*	108± 25*	42± 4
Cortical bone analysis
Mean thickness (mm)	0.096± 0.006*	0.120± 0.005	0.132± 0.004
Bone mineral density (mm/cc)	929± 30*	999±12	1038± 17
Bone mineral concentration (μg)	2.1± 0.2*	2.8± 0.1	3.3± 0.2
μMRI Femur (n=5 for each group)
Trabecular bone analysis
Surface-to-curve ratio (S/C)	4.11± 0.59*	3.79± 0.65*	6.24± 0.20
Erosion Index (EI)	1.14± 0.18*	1.11± 0.12*	0.69± 0.03

(−/− = PKBalpha/Akt1 knockout; +/− = heterozygote; +/+ = wild type; mean ± SD;

^*Significantly different from the wild type (+/+); P<0.05)

Trabecular bone of the right femur (Fig. 5c) was subsequently examined by μMRI, to validate the findings in the humerus. A major determinant of trabecular bone strength is orientation and is used for understanding the implications of osteoporotic bone loss in clinics (24). While CT is more sensitive to mineralized bone, μMRI shows particular sensitivity to the confinement of water within the developing trabecular bone. MRI microscopy of the femur depicted the trabecular bone within the epiphysis and metaphysis. Composite structural parameters (S/C and EI) were obtained by subjecting segmented volumes of the distal femoral metaphysis to the Virtual Bone Biopsy (VBB) processing algorithm. The femoral metaphysis of PKBalpha/Akt1 −/− mice, and notably also +/− mice, showed significantly more rod-like than plate-like morphology, which translates into decreased S/C, compared with +/+. An increased EI, which indicates more perforated trabecular plates, was observed in the trabecular network of PKBalpha/Akt1 null and heterozygote, while no difference was seen between heterozygote and knockout. These substantial changes in trabecular architecture of the metaphysis of the distal femur, in particular as they relate to topology of the network, were detected between PKBalpha/Akt1 −/− and +/− as compared to +/+ mice implicate reduced trabecular bone strength in null and heterozygote mice, suggesting a significant gene dosage dependent role for PKBalpha/Akt1 in bone development(Fig. 7; Table 1).

Discussion

Vascular development is critical for elongation of the long bones and thereby are an important factor in determination of the individual size. We have shown here that macromolecular DCE-MRI provides a sensitive method for detection of impaired vascular function in the humerus. We described deficient vascularization of the humeral metaphysis of the growth plate, in PKBalpha/Akt1 knockout (−/−) and heterozygote (+/−) mice, which could be detected in vivo by macromolecular DCE-MRI. In addition, as it is known that new bone formation follows angiogenesis, we showed in both PKBalpha/Akt1 +/− and −/− deficient metaphyseal trabeculae, probably due to a deficient supply of minerals and specialized cells.

Non-invasively quantifying vascular function and identifying vascular defects at the growth plate of long bones in living animals, gives access to important dynamic information that could not be obtained using classical methods, such as histology. Because of its high molecular size, biotin-BSA-GdDTPA acts as a blood pool agent and can only extravasate from leaky vessels. This improves the specificity for detection of small changes in blood volume and vascular permeability. Macromolecular and low-molecular weight DCE-MRI have been used in previous studies, as a valid and non-invasive method for semi-quantitive evaluation of microcirculation in bone pathologies with increased circulation such as solid tumors and bone marrow disease (20,25-27), but has not been described in evaluation of decreased circulation in bones, associated with genetic defects leading to developmental growth retardation.

In long bones, such as the humerus and femur, the main blood supply are the diaphyseal artery (entering through a single nutrient foramen; Fig. 7a), the metaphyseal, epiphyseal and periosteal arterial networks. In the growing long bone, leakiness of the blood vessels includes both the newly formed blood vessels near the growth plate, and also the sinusoids in the entire long bone (28). This permeability is evident in the behavior of the contrast agent during the first 15min. Unlike other tissues with permeable vasculature, lymphatics are not present in bone tissue (29) and thus contrast behavior is not characterized by lymphatic drain. Furthermore, the last branches of the nutrient artery to the metaphysis in growing bone loop very sharp and empty into a large sinusoidal system where the rate of blood flow is decreased. This slow flow at the metaphysis of growing bone, is illustrated in osteomyelitis in children, where bacteria in circulation can easily home at the metaphysis with slow blood flow (30). After 15min, the macromolecular contrast accumulated further in the bone. This accumulation is unique to the bone and can be explained by a complex interplay of continuous deposition in sinuoids, convection, and slow clearance of contrast agent.

In this work, macromolecular DCE-MRI proved to be sensitive for measuring impaired vascular function at the humerus of PKBalpha/Akt1 −/− and +/− mice. FBV, a measure for (micro)vascular density, in the humerus and its growth plate area was found to be significantly reduced in −/− and +/−mice. Furthermore, PS was reduced in −/− mice. The results were validated by fluorescence microscopy, revealing reduced localization of the BSA-ROX, an early vascular marker, distal from the growth plate, and also less MR contrast agent where newly formed metaphyseal blood vessels appeared leaky. After 15min, the contrast accumulates continuously, also in +/− and −/− mice, but to a lesser extend then in +/+ mice. These results indicate impaired vascular function in +/− and even more severe impairment in PKBalpha/Akt1 −/− mice.

Blood supply in the bone is necessary for sufficient deposition of minerals and provision of highly specialized cells for bone formation, and more particularly for trabecular bone formation near the growth plate (4). In PKBalpha/Akt1 −/−, osteopenia was previously described (14), but impaired bone formation in PKBalpha/Akt1 +/− mice has not been reported. Our results, of two separate long bones, using both μCT and μMRI, show significantly reduced trabecular bone formation in +/− as well. Cortical bone parameters in +/−, appeared not reduced compared to +/+ bones. This complies with the less severe vascular phenotype of +/− at the humerus. Dosage-dependent effects of PKBalpha/Akt1 on animal survival and development have been described previously only on the background of PKBgamma/Akt3 −/− mice (31).

Previously, growth retardation (12) and in utero lethality (11) of PKBalpha/Akt1 −/− mice were reported, and attributed independently to placental hypovascularity (32), or to osteopenia (13,14). PKBalpha/Akt1 is predominantly expressed in endothelial cells and is a central mediator in signaling of angiogenesis (33). Reduced angiogenesis at growth plates, loss of metaphyseal blood vessels and consequent reduction in trabecular bone formation, have been shown in studies of inhibition of VEGF signaling (4). VEGF expressed by hypertrophic chondrocytes mediates blood vessel invasion into the zone of hypertrophic cartilage (34) and this infiltration of endothelial cells boosts the functional differentiation of other cells particularly osteoblast and osteoclasts (9). Reduction in trabecular bone mass and defects in osteoblasts and osteoclasts have already been demonstrated in PKBalpha/Akt1 −/− mice (14). Moreover, a substantial reduction in monocyte-derived chondroclasts which are tartrate-resistant acid phosphatase (TRAP) positive cells, was previously reported in PKBalpha/Akt1 deficient mice (14), as was also seen in inhibition of VEGF signaling, indicating impaired recruitment of these specialized cells due to reduced neovascularization at the growth plate (4). A recent study (13), showed decreased VEGF immunostaining at the site of new bone formation, suggesting a possible role for PKBalpha/Akt1in angiogenesis. Nevertheless, no quantitative measurements of newly formed blood vessels were provided, and +/− and +/+ were taken together as control group for −/− mice.

In summary, we showed here significant vascular defects using the humerus, in particular of metaphyseal blood vessels, in both PKBalpha/Akt1 deficient and heterozygote mice. In addition, loss of a single allele of PKBalpha/Akt1 was sufficient to affect bone vessel density at the epiphyseal plate region, and (micro)vascular density of the long bone. We confirmed reduced trabecular and cortical bone formation in PKBalpha/Akt1 null mice. Moreover, we show that trabeculae were aberrant in growing PKBalpha/Akt1 heterozygote mice, underscoring the gene dosage dependent regulation of endochondral bone growth by PKBalpha/Akt1. MRI was instrumental for analysis and quantification of vascular function in the growing bone and for analysis of trabecular bone mineralization.

Abbreviations

−/−

gene deficient mice, both alleles of the gene were deactivated

+/−

heterozygous mice, one allele of the gene was deactivated

+/+

wild type mice, both alleles of the gene are present

biotin-BSA-GdDTPA

bovine serum albumin conjugated to biotin and Gd-DTPA

PKBalpha/Akt1

protein kinase B alpha, also known as Akt1

PKBgamma/Akt3

protein kinase B gamma, also known as Akt3

BSA-ROX

bovine serum albumin labeled with rhodamine

H&E

hematoxylin and eosin staining

PI3K

phosphatidyl-inositol 3-kinase, intracellular signal transduction enzyme

VEGF

vascular endothelial growth factor

fBV

fraction of blood volume

PS

permeability surface area product

BVF

bone volume fraction

VBB

Virtual Bone Biopsy

DTA

digital topological analysis

S/C

surface-to-curve ratio

EI

erosion index