Treatment with N- and C-Terminal Peptides of Parathyroid Hormone-Related Protein Partly Compensate the Skeletal Abnormalities in IGF-I Deficient Mice

Insulin-like growth factor-I (IGF-I) deficiency causes growth delay, and IGF-I has been shown to partially mediate bone anabolism by parathyroid hormone (PTH). PTH-related protein (PTHrP) is abundant in bone, and has osteogenic features by poorly defined mechanisms. We here examined the capacity of PTHrP (1–36) and PTHrP (107–111) (osteostatin) to reverse the skeletal alterations associated with IGF-I deficiency. Igf1-null mice and their wild type littermates were treated with each PTHrP peptide (80 µg/Kg/every other day/2 weeks; 2 males and 4 females for each genotype) or saline vehicle (3 males and 3 females for each genotype). We found that treatment with either PTHrP peptide ameliorated trabecular structure in the femur in both genotypes. However, these peptides were ineffective in normalizing the altered cortical structure at this bone site in Igf1-null mice. An aberrant gene expression of factors associated with osteoblast differentiation and function, namely runx2, osteoprotegerin/receptor activator of NF-κB ligand ratio, Wnt3a , cyclin D1, connexin 43, catalase and Gadd45, as well as in osteocyte sclerostin, was found in the long bones of Igf1-null mice. These mice also displayed a lower amount of trabecular osteoblasts and osteoclasts in the tibial metaphysis than those in wild type mice. These alterations in Igf1-null mice were only partially corrected by each PTHrP peptide treatment. The skeletal expression of Igf2, Igf1 receptor and Irs2 was increased in Igf1-null mice, and this compensatory profile was further improved by treatment with each PTHrP peptide related to ERK1/2 and FoxM1 activation. In vitro, PTHrP (1–36) and osteostatin were effective in promoting bone marrow stromal cell mineralization in normal mice but not in IGF-I-deficient mice. Collectively, these findings indicate that PTHrP (1–36) and osteostatin can exert several osteogenic actions even in the absence of IGF-I in the mouse bone.


Introduction
The insulin-like growth factor (IGF) system, formed by insulinlike peptides, their receptors and binding proteins, plays a central role in the regulation of cell growth and differentiation. Human homozygous loss-of-function IGF1 gene mutations cause intrauterine and postnatal growth failure and severe sensorineural deafness (OMIM 608747) [1,2]. Treatment with recombinant human IGF-I has been shown to improve short stature in patients with severe IGF-I deficiency [3], supporting the key role of IGF-I in skeletal development. Moreover, decrease in IGF-I production and/or activity has been suggested to contribute to age-related osteopenia and low bone formation [4,5]. Mice with a homozygous Igf1 gene deletion display a 30% size reduction and an aberrant bone phenotype with shortened femoral length and reduction in cortical bone size [6][7][8], and also sensorial impairment [9], as compared to wild type littermates. These bone changes are related to a decrease in both bone formation and bone resorption -with a low number of osteoblasts and osteoclasts-, and also a reduced capacity for osteoblastogenesis and osteoclastogenesis in the bone marrow of Igf1-null mice [6][7][8]. Thus, the observation of an increased trabecular bone volume in the proximal tibia of these mice was suggested to be a consequence of the IGF-I effect on osteoclast formation and/or activity at this skeletal site, which is absent in Igf1-null mice [6][7][8]. Mice with a homozygous deletion of the gene encoding the IGF-I high affinity receptor (Igf1r) show a delayed ossification in the cranial and facial bones, inner ear alterations and die shortly after birth [10,11]. Furthermore, partial deletion of the Igf1r gene causes postnatal growth retardation in humans [12]. IGF1R activation recruits insulin receptor substrates (IRS). Mice with homozygous deletion or spontaneous mutation in Irs1 show sensory alterations [13][14][15], severe bone growth impairment and low-bone turnover osteopenia [16,17]. In addition, gain-of-function mouse mutants of IGF-I binding proteins that reduce IGF-bioavailability also consistently show a low cortical and trabecular bone mineral density (BMD) and alterations in bone formation rates [18][19][20][21].
The bulk of current studies performed in rodents support the notion that the IGF system plays a paramount role in the bone anabolic actions of PTH [22]. Thus, neither Igf1-null nor Igf1r-null mice show the bone anabolic response triggered by transient administration of PTH in normal mice [23]. IGF1R in mature osteoblasts seems to be a critical PTH target for its skeletal actions [23]. Cells of the osteoblastic lineage are a rich source of PTHrelated protein (PTHrP), an important modulator of bone development and bone remodelling [24]. PTH and PTHrP interact with the same PTH type 1 receptor (PTH1R) in osteoblasts [25]. Similarly to PTH, intermittent administration of the N-terminal PTHrP fragment induces bone anabolic features in mice and humans [24][25][26][27][28][29][30]. Recent findings have also shown that the C-terminal PTH-unrelated region of PTHrP containing the osteostatin epitope (107-111) may also contribute to its osteogenic actions [24,[31][32][33][34][35]. Of interest in the frame of the present study, global deletion of Igf1 [6] or chondrocyte specific deletion of its receptor gene Igf1r [36] produces a severe phenotype which resembles that of PTHrP-null mice at birth [37], suggesting that PTHrP may signal through IGF-I during bone development. However, the putative contribution of IGF-I to the osteogenic features of the different domains of PTHrP is yet to be explored.
In the present study, we aimed to evaluate whether the response of mouse bone to PTHrP might be affected by the IGF-I status. Specifically, we determined and compared the skeletal effects elicited by PTHrP  or osteostatin administration to both Igf1-null mice and their wild type littermates, at the tissue, cellular, and molecular levels.

Mouse Genotyping and Functional Characterization
Heterozygous mice with a targeted disruption of the Igf1 gene were generated and maintained on a hybrid genetic background of MF1 and 129/Sv strains [10]. DNA extraction for genotyping was performed with the REDExtract-N-Amp TM Tissue PCR Kit (Sigma-Aldrich, St. Louis, MO), according to the manufacturer's instructions. PCR primers and conditions used were as previously reported [38,39]. The animals were fed a standard diet and drinking water ad libitum, and housed following the recommendations of Federation of European Laboratory Animal Science Associations. All animal experimentation was conducted in accord with Spanish and European legislation (EU directive 2010/63/ EU) and approved by the Animal Care and Use Committees of Spanish National Research Council (Consejo Superior de Investigaciones Científicas) and Instituto de Investigación Sanitaria-Fundación Jiménez Díaz.
Initially, young adult (2 and 4 month-old) Igf1 2/2 (Igf1-null), Igf1-heterozygous and Igf1 +/+ (wild type) mice were characterized by evaluating serum IGF-I levels, using a standard OCTEIA Rat/ Mouse IGF-I kit (IDS Ltd., Boldon, UK), according to the manufacturer's recommendations [9]. Non invasive tests of neural function, namely auditory brainstem response and sciatic nerve conduction velocity were also performed in these mice, as previously reported [14,39,40]. Briefly, auditory brainstem response was evaluated with a TDT System 3 TM workstation and the specific software SigGenRP TM and BioSigGenRP TM (Tucker Davis Technologies, Alachua, FL). Determination of sciatic nerve conduction velocity was measured by supramaximal stimulation (9 V, 2 mA, and 0.1 ms) at the sciatic notch and recording at the metatarsian region with a clip electrode. Latencies were measured in each case from the beginning of the stimulus to the first positive wave of the compound muscle action potential. The sciatic motor nerve conduction velocity was calculated dividing the measured latency by the distance from the sciatic notch to the clip electrode. As expected, Igf1-null mice showed undetectable IGF-I serum levels, a significantly reduced body weight, severe sensorineural deafness and a reduction in the sciatic nerve conduction velocity when compared to wild type mice. As expected, no differences related to age (within 2 and 4 months) were found in these parameters in each group of mice studied; therefore, pooled data are shown in Figure 1. Igf1-heterozygous mice showed a similar functional phenotype to wild type mice and were no further studied. Considering the aforementioned data, we selected two month-old Igf1-null and wild type mice for further studies.

Treatments
Eighteen mice of each Igf1-null or wild type genotypes were used. Groups of 6 mice (2 males and 4 females) of each genotype were treated with 80 mg/Kg/every other day of either PTHrP  or osteostatin subcutaneously for 2 weeks. The control group (3 males and 3 females) of each genotype received the same treatment with phosphate-buffered saline (PBS) vehicle. This time period is within the time frame used for initial characterization of these mice, as described above. Thus, differences observed during 2 weeks (period of study) are unlikely to be accounted for by age but by genotype and type of treatment. Since Igf1-null mice have a very poor postnatal survival rate (around 20%) [10], mice from both genders were used per genotype and type of treatment. We complied with the 3R (''replace, reduce, and refine'') experimental design recommendation aimed to reduce the number of experimental animals [41]. The number of mice in our study is in the range of those used in related studies [23,26,27,40,42]. The administered dose of PTHrP (1-36) is similar or lower than that represented by daily PTH administration in other mouse models with IGF-I action deficiency [23,42,43], and it has been shown to produce bone anabolism in rodents [26,28,[44][45][46]. A higher (in molar terms) osteostatin dose was selected, which was similar to that proven to exert anti-resorptive features in mice [47]. Mice tolerated well these treatments, and no secondary effects were observed. Animals were subjected to a 2-week treatment as described above and they were sacrificed 2 h after the last injection of each peptide or vehicle in order to analyze putative rapid changes in gene or protein expression. At the time of euthanasia, both tibiae and femurs were removed, and tissue from individual mice was obtained for histological analysis, total RNA or protein extraction.
Dual-energy X-ray Absorptiometry (DXA) and mcomputed Tomography (mCT) Analysis Bone mass changes were assessed in anesthetized mice by DXA (PIXIMUS; GE Lunar Corp., Madison, WI) [28,48]. Only wild type mice were evaluated since body weight of Igf1-null mice (,10 g) lies outside of the PIXIMUS automatic thresholds, as recommended by the manufacturer. Both BMD and BMC were assessed in the same region of interest in each skeletal compartment tested (total body, left femur, and L1-L5 vertebrae) at the start of the study and at the end of treatments using PIXIMUS software. Our animals were young adult and thus growing. Hence, in order to disclose putative bone mass differences between different treatments, BMD and BMC results were presented as percentage of changes as follows: 1006(BMD 14d -BMD 0d )/ BMD 0d , and 1006 (BMC 14d -BMC 0d )/BMC 0d (14 d and 0 d subscripts denoting the corresponding values at the end and the beginning of the study, respectively) in each experimental group.
Femurs were scanned using a high-resolution mCT system (Scanco Medical AG, Brüttisellen, Switzerland), with a voxel resolution of 10 mm 3 , an energy of 55 kVp, an intensity of 145 mA and an integration time of 200 ms. Thresholds applied were 392 mg Hydroxyapatite (HA)/cm 3 and 500 mg HA/cm 3 for trabecular and cortical bone, respectively. One hundred slices were evaluated in 1-mm of the femoral diaphysis (cortical bone); the number of slices tested in the metaphysis (trabecular bone) varied, given the different bone length in wild type and Igf1-null mice, consistently starting at 66% of the femur height down to the growth plate. The three dimensional microarchitectural properties of the cortical bone in the mid-diaphysis and trabecular bone in the distal metaphysis were assessed by using MFEM, R-10 (version V1.2) software (b-Cube AG, Schlieren, Switzerland). The following parameters were calculated: % bone volume/total tissue volume (BV/TV), bone surface (BS), trabecular thickness (Tb.Th), trabecular number (Tb.N), trabecular connectivity density (Conn. D), total area (T.Ar); cortical thickness (Ct.Th) and the polar moment of inertia (J).

Bone Histology
Mouse femoral samples were fixed in 4% p-formaldehyde in PBS and subsequently decalcified in Osteosoft (Merck, Darmstadt, Germany), dehydrated, and embedded in paraffin. Histological evaluation was carried out by Masson's staining on sagittal 3 mm sections from each mouse in 4 mice per experimental group. The growth plate width in wild type and Igf1-null mice was measured with NIH Image J software. Trabecular abundance was also calculated, and expressed as the trabecular number/mm in a bone area below the growth plate. Osteocytes were counted in 4 to 6 random x400-fields per sample in a cortical bone segment between the growth plate and the mid-diaphysis; the corresponding mean score value was normalized to the bone area of each sample. Osteoblasts, identified by their cuboidal shape and localization on bone surfaces and multinuclear osteoclasts (with 3 or more nuclei) [26,49], were counted in a 0.8-mm 2 area of the tibial metaphysis immediately below the growth plate. Evaluations were performed by 2-3 independent observers in a blinded fashion for each mouse.

Bone Marrow Stromal Cell (BMSC) Cultures
To obtain BMSCs, the bone marrow from both tibiae and femurs of either five Igf1-null or two wild type mice were pooled and cultured, as previously described [26,28]. BMSCs were seeded in a-minimum essential medium containing 15% heat-inactivated foetal bovine serum, 1% penicillin-streptomycin, and 2.5 mg/ml fungizone at a density of 1-2.5610 6 cells/cm 2 , onto 6-well plates in 5% CO 2 at 37uC. This culture medium supplemented with Figure 1. Phenotypic characterization of Igf1-null mice. IGF-I serum levels, body weight, as well as auditory brainstem response [hearing threshold, dB sound pressure level (SPL)] and sciatic nerve conduction velocity were measured in young wild type (WT; Igf1 +/+ ), heterozygous (Hz; Igf1 +/2 ), and Igf1-null (Igf1 2/2 ) mice, as described in Methods. Data from 2-and 4-month old mice of each genotype were pooled for statistical comparisons. Results are expressed as mean 6 SEM corresponding to the following number of mice (males and females, respectively): 7 and 4 (WT); 8 and 4 (Hz) and 2 and 2 (Igf1 2/2 ). ***p,0.001 vs WT mice. doi:10.1371/journal.pone.0087536.g001 50 mg/ml ascorbic acid and 10 mM b-glycerophosphate (osteogenic medium) was added at day 3 with or without PTHrP  or osteostatin (each at 100 nM). Half of the volume of the cellconditioned medium and peptide treatment was exchanged every other day. On day 16, matrix mineralization was determined by staining with 40 mM alizarin red S, pH 4.2, for 10 minutes, and measuring absorbance at 540 nm [28].

Gene Expression Analysis by Real Time PCR
Femoral samples of each Igf1-null and wild type mice were crushed under liquid nitrogen before total RNA extraction with Trizol (Invitrogen, Groningen, NL). For gene expression analysis by real time PCR, cDNA was generated from equal amounts of total RNA obtained from individual mice by reverse transcription (High Capacity cDNA Reverse Transcription Kit; Applied Biosystems, Foster City, CA). Thermal cycling and fluorescence detection was performed on an ABI PRISM 7500 or 7900HT system (Applied Biosystems). TaqMan MGB probes obtained from Assay-by-Design SM or TaqManH Gene Expression Assays were used (Applied Biosystems) for amplification of: i) osteoblastic genes including runt related transcription factor 2 (Runx2), osteocalcin (OC), osteoprotegerin (OPG), receptor activator of nuclear factor-kB ligand (RANKL), and Pth1r; ii) canonical Wnt pathway-related genes, Wnt3a, cyclin D1 (Ccnd1), and connexin43 (Cx43); iii) oxidative stress-related genes, catalase and ''growth arrest and DNA-damage-inducible 45 alpha'' (Gadd45); and iv) genes related to the IGF system and IGF-I signalling targets, including Igf1, Igf2, Igf1r, Irs2, tyrosine-protein phosphatase non-receptor type 1 (PTP1B) (Ptpn1), and the cell cycle activator forkhead box M1 (FoxM1). 18S ribosomal RNA or the mouse ribosomal phosphoprotein P0 (Rplp0) was used as endogenous control gene to normalize the expression data obtained. The relative quantification values (RQ) between Igf1-null and wild type mice (treated or untreated) were determined by the 2 2DDCt method, where DDCt = DC target gene -DC reference gene [50], and data were expressed as mRNA relative levels vs corresponding values in untreated wild type, as reported [38,49].

Western Blotting
Mouse intact tibiae from each experimental animal were homogenized in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 2 mM EGTA, 0.2% Triton X-100, 0.3% NP-40, 1 mM dithiothreitol and protease and phosphatase inhibitor cocktail (Sigma-Aldrich) as described [49]. Protein concentration was measured by bicinchoninic acid-based assay (Thermo Fisher Scientific, Rockford, IL) using bovine serum albumin (BSA) as standard. Analysis of protein expression was performed by Western blotting. Equal amounts of protein from individual mice were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis under reducing conditions and transferred to polyvinylidene fluoride or nitrocellulose (for sclerostin) membranes in a Bio-Rad Trans Blot according to the manufacturer's instructions. After incubation with a blocking solution, the membranes were probed overnight at 4uC with the following primary antibodies: anti-phospho (p)-AKT (Ser473); anti-p-p44/ 42 extracellular signal-regulated kinase (ERK) 1/2 or anti-p-p38 mitogen activated protein kinase (each at 1:1000 dilution), and anti-p44/42 ERK1/2 (Cell Signaling Technology; Danvers, MA), or anti-p38a (Santa Cruz Biotechnology; Santa Cruz, CA) (each at 1:1000 dilution) as loading controls, respectively; and antisclerostin antibody (R&D systems; Minneapolis, MN) (0.2 mg/ ml). Anti-a-tubulin antibody (Sigma-Aldrich) was used as loading control for the latter primary antibody. Antibodies were diluted in Tris-buffered saline with Tween containing 5% BSA for phosphorylation-specific antibodies or non-fat dried milk for the other antibodies. The membranes were washed and incubated with the corresponding peroxidase-conjugated secondary antibodies for 1 h at room temperature. Immunoreactive bands were visualized by enhanced chemiluminescence (GE Healthcare; Buckinghamshire, UK) using X-ray films, and the bands were quantified by densitometry with NIH Image J software.

Statistical Analysis
Statistical comparisons of hearing thresholds, NCV and IGF-I serum levels were performed with one factor ANOVA with post hoc Bonferroni or Tamhane test in case of homogeneous or nonhomogeneous variances according to Levene's test, respectively, using SPSS 19.0 (IBM SPSS statistics). Statistical analysis of changes in the expression of IGF system and related genes, assessed by real time PCR, was performed using the Integromics Real Time StatMiner software package (http://www.integromics. com/genomics-data-analysis/pcr-analysis). Statistical significance of IGF system protein expression levels as well as differences in matrix mineralization and bone cell numbers were analyzed by ANOVA with post-hoc Bonferroni test. Other statistical comparisons were done by Kruskal-Wallis test followed by Mann-Whitney test. Results were expressed as mean 6 SEM. p,0.05 was considered significant.

Changes in Bone Structural Parameters Elicited by PTHrP (1-36) and Osteostatin in Igf1-null and Wild Type Mice
Igf1-null mice showed a significant decrease in femur length (Figure 2A, left), together with a dramatic decrease in the width of the growth plate and reduced trabecular number in the femoral metaphysis, compared to wild type mice (Figure 2A, right). Moreover, using mCT evaluation, a general alteration was observed in various parameters in both trabecular and cortical compartments in the femur of Igf1-null mice compared to wild type mice (Tables 1 and 2 and Figure 2B). No significant differences related to PTHrP peptide treatment were observed in body weight in each genotype studied at the end of the study. These values were (g; mean6SEM; n = 6): 26 Igf1-null+osteostatin). Bone mass differences in wild type mice throughout the study were analyzed by calculating the percentage of change of BMD and BMC values for each mouse, as stated in Materials and Methods. By using this approach, we were able to detect a significant (p,0.05 or lower) increase in these parameters in the femur (but not in the total body or the vertebrae) after treatment with each PTHrP peptide tested. These % values were (mean 6 SEM corresponding to 3 males and 3 females for untreated mice, and 2 males and 4 females for PTHrP-treated mice): 0.161.2 (wild type, WT); 5.962.2 [WT+PTHrP (1-36)]; and 8.161.4 (WT+osteostatin) (DBMD, g/cm 2 ); or 0.162.5, 9.263.1, 11.862.7 (DBMC, g), respectively.
By using mCT analysis of the distal femur, differential changes elicited by each PTHrP peptide tested were observed in each IGF-I scenario. Thus, in wild type mice, both PTHrP (1-36) and osteostatin were similarly effective in stimulating cortical parameters, namely T.Ar., Ct.Th and J (Table 1), and also Tb.Th (Table 2) at the femoral metaphysis. Meanwhile, administration of PTHrP  to Igf1-null mice improved all the trabecular parameters evaluated even above control values, but osteostatin treatment was significantly less efficient in this bone compartment. Moreover, neither PTHrP peptide tested showed efficacy to  normalize the altered cortical parameters determined in the femur of these animals (Tables 1 and 2). These alterations in the distal femur of mice with different Igf1 genotype and the osteogenic actions of PTHrP (1-36) and osteostatin are depicted by corresponding mCT images ( Figure 2B).

Comparative Effects of PTHrP (1-36) and Osteostatin on the Expression of Bone Related Factors in Igf1-null and Wild Type Mice
In the femur of wild type mice, treatment with either PTHrP peptide caused a similar increase in the expression of the early osteoblast differentiation transcription factor Runx2, but not in that of the late osteoblast differentiation marker OC (Figure 3). OPG/ RANKL ratio is considered to be a major modulator of bone resorption and bone remodelling [51]. In these mice, only osteostatin treatment enhanced the OPG/RANKL mRNA ratio, by increasing OPG and decreasing RANKL gene expression (Figure 3), coherent with its anti-resorptive features [47]. On the other hand, treatment with PTHrP (1-36) but not osteostatin increased the gene expression levels of catalase and Gadd45 two oxidative stress-associated genes [52] (Figure 3).
In the femur of Igf1-null mice, Runx2 gene expression levels and the OPG/RANKL mRNA ratio were increased three times over the corresponding levels of the wild type mice, and were unchanged by either peptide treatment ( Figure 3). This observation is consistent with the reported low status of osteoblastic maturation in Igf1-null mice [23]. In these mice, PTHrP (1-36) administration increased OC mRNA levels, and reversed the down-regulation of catalase and Gadd45 (Figure 3). Pth1r mRNA expression was unchanged in these mice (data not shown), as reported in another Igf1-null mouse mutant [23].
Activation of the canonical Wnt pathway is an important mechanism to foster bone formation [53]. We found that the expression of Wnt3a, Ccnd1 and Cx43 was strongly decreased in Igf1-null mice. This decrease was partially compensated for by PTHrP treatment (Figure 4A). Sclerostin is an inhibitor of the canonical Wnt pathway produced by osteocytes, which acts as an important modulator of bone remodelling [54]. We found that sclerostin protein levels were diminished in Igf1-null mice; and administration of either PTHrP peptide prevented in part this decrease ( Figure 4B). These changes were not related to those in the number of osteocytes in the tibia of Igf1-null mice ( Figure 4C).
We next evaluated whether differences in the observed osteogenic action of each PTHrP peptide according to the mouse genotype were reflected in corresponding changes in osteoblasts and osteoclasts in the mouse tibia. Igf1-null mice were found to display a lower amount of both cell types per bone tissue area (mm 2 ), compared to wild type mice ( Figure 5A and B).
Furthermore, the abundance of osteoblasts significantly increased after PTHrP administration in the latter mice but not in Igf1-null mice ( Figure 5A). On the other hand, either peptide was also effective in decreasing the number of osteoclasts only in wild type mice ( Figure 5B).
For assessing whether PTHrP might display cell autonomous actions in an IGF-I deficient environment related to its observed osteogenic actions in Igf1-null mice, we used ex vivo BMSC cultures from either these mice or wild type mice. BMSCs from Igf1-null mice had a reduced matrix mineralization capacity compared to those from wild type animals. In addition, matrix mineralization significantly increased upon treatment with osteostatin in these cell cultures only from the latter mice ( Figure 5C).

PTHrP (1-36) and Osteostatin Modulate IGF System Expression and IGF Signalling in the Long Bones of Igf1null Mice
We next evaluated several components of the IGF system and downstream signalling pathways, which might have been targeted by the PTHrP peptides to compensate for the absence of IGF-I. Basal expression of Igf2, Igf1r and Irs2 was found to be increased in the femur of Igf1-null mice compared to wild type mice, possibly as a response to IGF-I deficiency. Interestingly, this was not the case for the IGF1R inhibitory tyrosine phosphatase Ptpn1 gene expression, which remained unchanged. The cell cycle activator FoxM1 gene was also increased in Igf1-null mice. Of note, the expression of these genes further increased in Igf1-null mice after treatment with either PTHrP peptide ( Figure 6A). Neither PTHrP peptide affected Igf1 mRNA expression in the femur of wild type littermates (data not shown).
Upon IGF1R activation, mitogen activated kinase-ERK1/2 and phosphatidylinositol-3 kinase/AKT downstream pathways are activated to promote cell proliferation and survival, respectively; whereas the pro-inflammatory p38a kinase pathway becomes inactivated [1,38]. We here examined the status of these signalling pathways in the tibia of Igf1-null mice, and the effect of PTHrP administration to these mice. AKT activation was significantly decreased in the Igf1-null mouse tibia, regardless of PTHrP peptide treatment ( Figure 6B). On the other hand, there were no significant changes in ERK1/2 and p38a phosphorylation in Igf1null mice compared to wild type mice. Together with FoxM1 gene expression data, these results suggest that cell proliferation is not severely compromised and could be maintained by the concerted actions of IGF-II and IGF1R in Igf1-null mice. Treatment with PTHrP (1-36) increased both p-ERK1/2 and p-p38a levels in the Igf1-null mouse tibia. In contrast, in these mice, there was no evident change in the former levels but a reduction in the latter levels upon osteostatin treatment ( Figure 6B). Discussion IGF-I-deficient mice have dramatic growth retardation, showing smaller bones than those of normal mice [1,4]. In this study, young adult mice with Igf1 gene deletion showed significant alterations in bone mass and bone structure at both cortical and trabecular compartments. Previous studies in IGF-I deficient mice of a different background have shown a decrease in cortical bone formation but an increase of several trabecular parameters in the tibia [6,8,23]. It has been hypothesized that bone regional differences in response to IGF-I deficiency might be a consequence of the dual effect of IGF-I on both osteoblastogenesis and osteoclastogenesis [4,6,55].
Differences have also been reported for the anabolic action of intermittent PTH or PTHrP (1-36) treatment on trabecular and cortical bone in mice and humans [30,56,57]. Here, we show that intermittent PTHrP  or osteostatin treatment for two weeks in wild type mice showed a similar efficacy at increasing bone mass related to enhancing various cortical parameters but not the majority of trabecular parameters in the femur. In this regard, a previous study has shown that the anabolic effect of PTH was mainly observed in cortical bone, whereas BV/TV was decreased Figure 3. Osteoblast-and oxidative stress-related genes in wild type and Igf1-null mice treated or not with PTHrP. Gene expression (assessed by real time PCR) of Runx2, OC, OPG, RANKL, catalase and Gadd45 in the femur of wild type (WT) and Igf1-null mice, treated or not with PTHrP (1-36) (Nt) or osteostatin (Ost). The variation coefficient of vehicle-WT (control normalized to 1) value was consistently ,30% in each case. This variation was taken into consideration for statistical analysis using Kruskall-Wallis test followed by Mann-Whitney test. Values are mean 6 SEM corresponding to 2 males and 4 females for PTHrP-treated mice or 3 males and 3 females for vehicle-treated mice of each genotype, respectively. *p,0.05; **p,0.01 vs vehicle-treated WT value; # p,0.05; ## p,0.01 vs vehicle-treated Igf1-null mice. doi:10.1371/journal.pone.0087536.g003 at trabecular level, after 2 week-treatment in male CD1 mice [23]. Moreover, gender differences in PTH bone anabolism have been reported in this mouse strain [58]. Normal male mice on a mixed genetic background of FVB/N, C57Bl/6J and 129Sv also showed a stronger PTH response of cortical bone than trabecular bone after 4 week-treatment [57]. Thus, although anabolic effects of PTH have been reported in both genders of different mouse strains [23,57,58], it cannot be ruled out that the osteogenic effect of PTHrP reported here might have been attenuated by combining both genders. A clear anabolic action of PTH in mouse trabecular bone has been reported using a prolonged treatment and/or higher PTH doses than those used here for PTHrP [44,57,59]. The observed increase in Tb.Th. by PTHrP administration in the femoral metaphysis of wild type mice could be explained by a possible (yet uncharacterized) effect of these peptides on lining cells as occurs with PTH [60]. Alternately, a putative anti-resorptive action of PTHrP peptides might contribute to this increase, a mechanism supported by our histology data and previous observations [26][27][28][29]35,47]. Further studies are needed to confirm these hypotheses. These findings suggest that differences in mouse strains and/or peptide administration regimes highly influence the bone anabolism achieved by PTH and presumably PTHrP peptides.
Our data indicate an IGF-I dependence for the osteogenic effects of PTHrP peptides on the cortical compartment of the mouse femur. In the present study, the administered doses of PTHrP peptides are higher than those used in a recent clinical study showing the bone anabolism of PTHrP  in postmenopausal women [30]. We are also aware of the limitation represented by using these high doses for reaching conclusions on the physiological relevance of the present findings. In any event, and consistent with our results, PTH doses in the range of those used here for PTHrP have been reported to be ineffective in the cortical long bones of mice with global IGF-I or osteoblastic IGF1R deficiency [23,61]. Also of interest, PTH was shown to be ineffective in cortical bone, but not in trabecular bone, in mice with liver-specific IGF-I deletion [62]. The low resorptive activity related to IGF-I deficiency has been suggested to account for the different PTH response of cortical and trabecular bone [6,23]. In this respect, the reduced number of trabecular osteoclasts in the Igf1-null mouse tibia suggests that such a mechanism might facilitate the disclosure of an anabolic action of PTHrP peptides in the trabecular femur of Igf1-null mice. Increased OPG mRNA levels (without changes in those of RANKL), which are likely to correlate with those of the corresponding protein [63][64][65], were also detected in these mice. This further supports the notion that a deficit of osteoclastogenesis occurs in Igf1-null mice.
Gene expression analysis also suggests the existence of an osteoblast maturation deficit in the absence of a pro-oxidative stress scenario in the femur of osteopenic Igf1-null mice. PTHrP (1-36) (but not osteostatin) treatment increased OC gene expression, suggesting it acts toward correcting the altered osteoblast differentiation in these mice. A similar efficacy of PTH in promoting the expression of osteoblast differentiation markers (including OC) in the femur of another Igf1-null mouse model has been previously reported [23]. However, the anabolic  Rplp0 expression levels were used as endogenous housekeeping control gene. Data correspond to 2 males and 3 females for PTHrP-treated mice or 3 males and 2 females for vehicle-treated mice of each genotype, respectively, and were calculated as log10RQ and represented as relative levels over corresponding WT-vehicle value in each case. Adjusted p-values were calculated with StatMiner software, and were considered significant when p,0.05. (B) Levels of p-ERK1/2, p-AKT and p-p38a were measured by and catabolic effects of PTH on the proximal tibia of wild type mice were absent in Igf1-null mice [23]. This apparent discrepancy was explained by the suggestion that IGF-I might not be essential for proximal events of osteoblast activation (namely, increased expression of some osteoblastic genes) by PTH. In addition, components of the canonical Wnt pathway were found to be affected in the long bones of Igf1-null mice, further underpinning the existence of an altered bone formation and remodelling in these animals. The observed decrease in sclerostin protein expression (without changes in osteocyte density) in Igf1-null mice indicates a diminished osteocyte function. In this respect, targeted disruption of Igf1 in mouse osteocytes results in poor Wnt pathway activation causing a marked impairment of bone development and in the bone anabolic response to loading [66,67]. Our data indicate that administration of PTHrP  or osteostatin partially recovered Wnt pathway activation in these mice. In fact, PTHrP (1-36) and the osteostatin-related PTHrP (107-139) peptide have recently been reported to target this pathway in osteoblastic cells in vivo and in vitro [48,68], related to their osteogenic action in diabetic mice with low bone turnover osteopenia [28,69].
A previous report has shown that primary calvaria osteoblasts isolated from mice with partial deletion of Igf1 showed decreased proliferation [70]. In the present study, the long bones of Igf1-null mice displayed less trabecular osteoblasts, and BMSCs from these mice had an impaired mineralization capacity. This was similar to previous findings in these cultures from mice with selective deletion of Igf1r in mature osteoblasts, which failed to respond to PTH [61,71]. In line with this observation, BMSCs from Igf1-null mice showed a lack of osteogenic differentiation response to PTHrP (1-36) or osteostatin in vitro; although this response occurred for osteostatin in BMSCs from wild type mice. However, PTHrP (1-36) was ineffective in this respect in the latter mice, in contrast to a previous report using PTHrP  and rat BMSCs [72]. Differences in either the N-terminal PTHrP peptide sequence or species (rats and mice) between this study and the present study might explain these discrepancies. Anyhow, our findings support a role for IGF-I in the action of PTHrP on bone marrow osteoprogenitors.
Bone cells synthesize IGF-I and IGF-II [22]; both factors bind the tyrosine kinase receptor IGF1R although with different affinities [1,73,74]. Upon this receptor activation, intracellular pathways are activated targeting important cell processes such as proliferation, survival, differentiation, inflammation and stress. Feedback signalling negatively regulates this activation by means of phosphatases and proteases [73]. We here show an increased gene expression levels of Igf2 and Igf1r concomitant to IGF-I deficiency in the mouse femur. Interestingly, an increased expression of docking Irs2 was also detected, whilst that of Ptp1n was unchanged. In this signalling scenario, ERK activation is maintained whereas that of AKT was decreased, which suggests that cell survival but not cell proliferation would be compromised in the bone tissue of Igf1-null mice. These data are in contrast to those obtained in Igf1-null cochlea, retina and brain, which do not exhibit an increase in the basal levels of either IGF-II or IGF1R or its substrates [9,38,39]. In the cochlea, stress pathways are strongly activated; in contrast, p38a phosphorylation levels were unchanged in the bone tissue of Igf1-null mice. Our findings of differential activation of IGF-I signalling mediators between the cochlea and bone of Igf1-null mice suggest that IGF-I deficit causes specific alterations in target tissues [4,74].
Expression levels of Igf2, Igfr1 and Irs2 were further increased in the femur of Igf1-null mice after treatment with each PTHrP peptide. Although this treatment failed to affect Igf1 mRNA levels at this skeletal site in wild type littermates (data not shown). In contrast, PTH treatment for a longer period (4 weeks) increased bone IGF-I content in rats [75,76]. Our data in Igf1-null mice also indicate that PTHrP (1-36) was effective in promoting ERK1/2 and p38a phosphorylation, suggesting that these pathways could be effectively modulated through IGF-II/IGF1R. On the contrary, the decreased p-AKT levels in these mice were not normalized after treatment with either PTHrP peptide. Therefore, PI3K/AKT pathway activation by PTHrP in bone seems to be IGF-I-dependent.
These aggregated findings confirm and extend the reported skeletal defects of Igf1-null mice using other mouse strains. In addition, our data in mice on a hybrid MF1/129/Sv genetic background support the notion that PTHrP (1-36) and osteostatin can exert osteogenic actions even in the absence of IGF-I.