Genetic studies in mice have demonstrated that deletion of many immediate early genes such as the AP-1 transcription factor constituents c-Jun, JunB, and Fra-1 results in a lethal phenotype [1–3], thereby underscoring their critical role in embryonic development. However, deletion of other immediate early genes, including c-Fos and JunD, has produced viable mice which presented with, at the time of discovery, unexpected bone abnormalities [4, 5]. Specifically, mice deficient in C-FOS suffer from pronounced osteopetrosis because of a lack of bone-resorbing osteoclasts [5], while animals lacking JUND have high bone mass due to enhanced bone formation [4]. Another immediate early gene potentially involved in postnatal bone biology is Egr-1, also known as NGFI-A [6], zif268 [7], krox24 [8], and TIS8 [9]. This transcription factor acts independently of AP-1 and preferentially binds to a GC-rich sequence, 5′-GCGGGGGCG-3′ [10–12], which has been located within the promoter region of a number of target genes important for proliferation, differentiation, growth control, tumor progression, inflammation, and apoptosis [13–25]. Despite the fact that EGR-1 regulates a broad array of genes in vitro, deficiency of EGR-1 in mice manifests in no visible phenotype and does not alter normal gross appearance, growth rate, and cellular differentiation [21]. However, there is evidence that in stromal cells from these mice a physical interaction between EGR-1 and SP-1 acts as a negative regulator of M-CSF gene expression [26]. Because M-CSF is important for osteoclast formation, this suggests increased osteoclast activity in Egr-1 −/− mice. Therefore, it is reasonable to hypothesize that EGR-1 functions as a regulator of skeletal properties. Because little is known about the role of EGR-1 in postnatal bone, this study assessed bone structural parameters and composition in both male and female mice. Analyses were carried out on appendicular long bones typically used for mouse phenotype analysis. Additionally, rib, an axial flat bone, was assessed to derive data from a distant skeletal site.

Materials and Methods

Animal Model

Heterozygous breeding pairs of the Egr-1 −/− mouse model 2013 (B6 background) were originally purchased from Taconic (Hudson, NY). Wild-type (wt, +/+) and Egr-1 −/− mice were derived from an in-house colony through a het/het breeding scheme and identified by standard genotyping procedures using polymerase chain reaction. Animal work was performed in accordance with the Institutional Animal Care and Use Committee. All animals for this study were offspring of the established mouse colony. Having reached the age of 12–14 weeks and prior to death, the body weight of animals was measured using a digital scale.

Tissue Preparation

All mice were killed using CO2 inhalation. Femurs were harvested by disarticulation of the femur in the hip joint to separate the lower limb from the torso. The patella tendon was cut horizontally and the tibia was separated from the femur. Subsequently, all surrounding muscle was removed. For rib tissue preparation a longitudinal incision was made along the thoracic spine, thereby exposing the dorsal aspect of the thorax. Ribs were counted from caudal to cranial to identify an eighth rib. For harvesting, tissue was dissected with scissors. Subsequently, all surrounding skeletal muscle was removed. For initial bone-length measurements, a manual caliper was utilized.

Radiography and Micro-Computed Tomography

A total of six male and seven female Egr-1 −/− bones were studied per group. The same numbers of age- and sex-matched wt bones from littermates served as controls. For radiography and subsequent micro-computed tomography (micro-CT), tibia, femur, and eighth rib were used. Radiography was performed with the MX20 Faxitron (Lincolnshire, IL) system at 26 kVp and 10 s with automatic exposure control. All bones were stored in 70% ethanol at 4°C until subjected to micro-CT analysis. Three-dimensional (3D) morphology of trabecular and cortical bones was determined within a 1.2-mm region of interest (ROI) defined at the distal epiphysis of the femurs starting 100 μm from the growth plate, a 1-mm ROI at the mid-diaphysis, and 3 mm of the middle section of the eighth rib using a Scanco μCT 35 (Scanco Medical, Bassersdorf, Switzerland) desktop system. All bones were scanned in 70% ethanol. A 6-μm voxel size, 55 kVp, 0.36° rotation step, 180° angular range, 400-ms exposure, and three averaged frames per view were used. The Scanco μCT software was used for 3D reconstruction, evaluation, and viewing of images. After 3D reconstruction, volumes of interest were segmented and analyzed using a global threshold of 0.4 g cm−3 for trabecular bone and 0.6 g cm−3 for cortical bone. In addition to bone mineral density (BMD, mg ml−1), bone morphometric measurements, including bone volume fraction (BV/TV, %), surface to volume ratio (BS/BV, mm−1), trabecular thickness (Tb.Th, mm), trabecular number (Tb.N, mm−1), trabecular separation (Tb.Sp, mm), and connectivity density (Conn.D, mm−3), were calculated for trabecular bone. For cortical and rib bone assessment, BMD, BV/TV [27], cortical thickness (Th, mm), maximum (I max, mm4) and minimum (I min, mm4) moments of inertia, and bone area (BA, mm2) were measured.

Fourier Transform Infrared Spectroscopic Imaging

Three male and three female animals form each genotype (−/− and wt) were randomly selected from the same animal cohort that was used for micro-CT, and the contralateral femur was prepared for FTIRI. All specimens were fixed with an alcohol gradient and embedded in polymethyl methacrylate. Embedded specimens were cut into 2-μm-thick sections using a microtome (SM2500; Leica, Bannockburn, IL) and mounted on barium fluoride infrared windows (SpectraTech, Hopewell Junction, NY). Sections were analyzed by FTIRI using a Spotlight 300 Imaging system (Perkin-Elmer Instruments, Waltham, MA). Based on a light microscopic micrograph of each section, femoral trabecular and cortical ROIs approximately corresponding to the micro-CT ROIs were scanned from 2000 to 800 cm−1 at a spectral resolution of 4 cm−1 and a spatial resolution of 6.25 μm as described elsewhere [28]. Because of possible variation in section thickness, we measured only relative spectroscopic parameters that were characterized before [29]. Briefly, those parameters included the mineral-to-matrix ratio, crystallinity, carbonate-to-phosphate ratio, and (collagen) cross-linking, which correlate with relative bone mineral content, mineral crystal size and perfection, carbonate substitution in hydroxyapatite, and collagen maturity, respectively [30]. After spectral subtraction of the embedding medium followed by spatial and spectral masking of the background in every scanned field, the spectroscopic results were expressed as mean values and standard deviations of the pixel distributions; and corresponding color-coded images were generated using ISYS software (Olney, MD). In addition, heterogeneity of FTIRI parameters was assessed by the full width at half-maximum calculated from the pixel histograms for each individual image [31]. Means and standard deviations were averaged for multiple sites in each animal for areas of cortical and trabecular bone.

Mechanical Testing

Mechanical properties of femurs were evaluated by three-point bending tests for wt (n = 6) and Egr-1 −/− (n = 6) bones. Tests were conducted at room temperature on a precision electromagnet-based load frame (EnduraTEC ELF 3200; Bose, Minnetonka, MN). Femurs were tested such that the anterior side was loaded in compression and the posterior side in tension. The posterior surface was placed on the lower supports, which were set 7.61 mm apart for both wt and Egr-1 −/− bones. Load was applied at 0.1 mm s−1 until failure occurred, and data were recorded at 100 data points/s. Structural mechanical properties that are dependent on geometry were measured and included max load (N), failure load (N), stiffness (N mm−1), max moment (N mm), and failure moment (N mm).

Statistical Analysis

Statistical analyses were carried out using a two-sample t-test, assuming equal variance; P < 0.05 was considered significant.

Results

Mice Deficient in EGR-1 Present with a Reduction in Body Weight

Male Egr-1 −/− mice measured a 13.5% reduction in body weight compared to wt male animals (Fig. 1a). A 13.7% weight reduction was also recorded for female Egr-1 −/− mice (Fig. 1a). Comparing the two sexes, female Egr-1 −/− mice were 11.5% lighter than male Egr-1 −/− mice, an expected finding that reflects the natural weight difference between female and male wt mice (Fig. 1a). Post mortem radiographic analysis was used for initial phenotype screening. No obvious differences were detected between knockout and wt femurs or ribs of either sex (Fig. 1b). A similar observation was made for tibias (data not shown). Assessment of bone length also did not reveal any significant difference between knockout and wt femurs or ribs in either male or female mice (Fig. 1c).

Fig. 1
figure 1

Weight reduction and normal bony gross appearance in Egr-1 −/− compared to wt (+/+) mice. a Animal weight at death. b Bone anatomy captured by radiography. c Overall length of excised bones was evaluated in male and age-matched female animals

Full size image

Reduced BV and BMD in Trabecular Bone of Egr-1 −/− Male Mice

To determine the consequence of EGR-1 deficiency on the mouse skeleton, we first performed bone measurements using micro-CT. A significantly (43%) reduced BV/TV was detected in femoral trabecular bone in male Egr-1 −/− mice compared to male wt animals (Fig. 2a). This was also the case for tibial trabecular bone (data not shown). Along with the decrease in BV/TV, a statistically significant 4% lower BMD was observed in trabecular, but not cortical, bone of male Egr-1 −/− mice (Fig. 2c, d). In contrast to male mice, no significant reduction in BV/TV or BMD was measured for trabecular or cortical bone in female mice (Fig. 2a–d). Planes roughly located within the ROI examined by micro-CT were analyzed on the molecular level utilizing FTIRI to detect changes in mineral-to-matrix ratio, carbonate-to-phosphate ratio, crystallinity, or cross-linking. There was no statistically significant change in any of these parameters between Egr-1 −/− and wt femur trabecular (Table 1) or cortical (Table 2) bone. For each parameter, we then compared the signal distribution (heterogeneity) within each ROI by visually inspecting the corresponding images and found a modest reduction in heterogeneity of the mineral-to-matrix ratio in Egr-1 −/− compared to wt cortical bone of the female femur (Supplementary Fig. 1). This prompted us to quantify signal heterogeneity for all parameters (not shown). A statistically significant decrease in heterogeneity of 16% (P = 0.04) was only observed for the mineral-to-matrix ratio in female Egr-1 −/− femur cortical bone.

Fig. 2
figure 2

Osteopenic bone characteristics in male Egr-1 −/− mice. Micro-CT measurements are presented and expressed as mean ± standard deviation. Statistical comparisons were made for each knockout (−/−) and wild-type (+/+) comparison, but only statistically significant (*P < 0.05) differences are indicated

Full size image
Table 1 FTIR parameters ± standard deviation for trabecular bone
Full size table
Table 2 FTIR parameter ± standard deviation for cortical bone
Full size table

Sex-Specific Alterations Cause Low Trabecular BV in Egr-1 −/− Mice

To investigate the origin of the reduction of BV/TV in trabecular bone, a more detailed micro-CT analysis of the trabecular morphology was conducted (Fig. 3). Trabeculae from male Egr-1 −/− mice showed a significant 17% reduction in number (Fig. 3a), associated with a 23% larger separation (Fig. 3c). As a result, the density of trabecular connections was also reduced (Fig. 3e). Additionally, thinner trabeculae were noted, but this observation was not significant (Fig. 3b). Trabecular bone from female Egr-1 −/− femurs was chiefly characterized by an 11% loss in thickness (Fig. 3b), which correlated to a 7% increase in BS/BV ratio (Fig. 3d). Although statistically not significant, there was a tendency for compensation as indicated by an increase in trabecular number (Fig. 3a) and a decrease in separation (Fig. 3c).

Fig. 3
figure 3

Trabecular bone morphology in EGR-1-deficient bones is sex-dependent. Micro-CT measurements are presented and expressed as mean ± standard deviation. Statistical comparisons were made for each knockout (−/−) and wild-type (+/+) comparison, but only statistically significant (*P < 0.05, **P < 0.001) differences are indicated

Full size image

Functional Abnormalities in the Cross-Sectional Geometry of Cortical Egr-1 −/− Bone

To assess the basis of the reduction of BV/TV in femur cortical bone, an extended micro-CT analysis of the cortical bone geometry was carried out (Fig. 4a–d). Statistically significant reductions of 10, 29, and 17% were seen for bone area in females and the minimum moment of inertia in males and females, respectively. Because the observed changes in cross-sectional geometry can influence the biomechanical properties of bone, three-point bending to failure was conducted on male Egr-1 −/− femurs, which showed the most pronounced reduction in the minimum moment of inertia. Testing revealed statistically significant reductions of 28, 42, 33, 28, and 42% in maximum load, failure load, stiffness, maximum moment, and failure moment, respectively (Fig. 4e–i). Hence, in male femurs, the loss of EGR-1 produced a change in cross-sectional geometry that led to reduced bending stiffness and strength during three-point bending.

Fig. 4
figure 4

Abnormalities in the cross-sectional geometry of cortical Egr-1 −/− bone. a–d Micro-CT measurements are presented and expressed as mean ± standard deviation. e–i Mechanical properties of the male femur as determined by three-point bending. Statistical comparisons were made for each knockout (−/−) and wild-type (+/+) comparison, but only statistically significant (*P < 0.05, **P < 0.001) differences are indicated

Full size image

Alterations in Female Rib Cortical Bone Differ from Femur Cortical Bone

In rib bone, micro-CT revealed 8 and 14% reductions in BV/TV in male and female Egr-1 −/− animals, respectively (Fig. 5a), compared to no significant change in femur cortical bone (Fig. 2b). Similar to femurs, however, no alteration in BMD was observed in either male or female ribs (Fig. 5b). With an 8% decrease in males compared to 17% in females, thickness was less affected in male ribs (Fig. 5c). A 20% reduction in bone area was seen in female, but not in male, Egr-1 −/− mice (Fig. 5d). Cross-sectional geometry in male Egr-1 −/− mice was not altered compared to wt mice (Fig. 5e, f), while female Egr-1 −/− animals had a 21% decreased minimum moment of inertia (Fig. 5f). Thus, compared to femur cortical bone, rib cortical bone showed decreased bone mass and, in female animals, distinct geometrical changes.

Fig. 5
figure 5

Distinct osteopenic traits in rib bone. Micro-CT measurements are presented and expressed as mean ± standard deviation. Statistical comparisons were made for each knockout (−/−) and wild-type (+/+) comparison, but only statistically significant (*P < 0.05, **P < 0.001) differences are indicated

Full size image

Discussion

In this study, we used a genetic approach to investigate the hypothesis that EGR-1 functions as a regulator of the postnatal skeleton. Bone properties were compared between wt and Egr-1 −/− mice at approximately 3 months of age. This time point was chosen because it permitted comparison to relevant published data [32]. Further, a single–time point examination freed resources for in-depth measurements at multiple skeletal sites in both male and female animals. Mice deficient in EGR-1 were smaller than wt mice, but femurs, tibias, and ribs presented with a normal gross appearance (Fig. 1b, c, and unpublished data). Across these three bone types, a net osteopenic phenotype was observed, as characterized by a reduction in BV/TV in all male and female bones except female trabecular bone (Figs. 2a, b and 5a; see supplementary tables). Further, BMD was reduced in all male and female bones except male rib bones (Figs. 2c, d and 5b; see supplementary tables). This observation is consistent with a previous report by Cenci et al. [32], who measured BMD over the hind limb of Egr-1 −/− and wt mice of comparable age and found a reduction in Egr-1 −/− animals. They also described an increase in both excretion of deoxypyridinoline and serum levels of osteocalcin in Egr-1 −/− mice [32], potentially suggestive of a bone turnover defect. However, there is no stringent correlation between detection of circulating bone turnover markers and a bone phenotype [33], perhaps, at least in part, because of a temporal or spatial divergence. Thus, in future studies it will be important to accurately compare Egr-1 −/− and wt mice for bone-turnover abnormalities using dynamic histomorphometry. Although we did not directly measure cellular markers of bone turnover, the use of FTIRI [29] permitted assessment of bone properties at the molecular level. This has been shown to be a sensitive method [34] for detection of bone-turnover abnormalities in rodent [35, 36] and human [37] bones. Our data did not reveal any difference in mineral-to-matrix ratio, carbonate-to-phosphate ratio, crystallinity, or cross-linking, which is a measure of collagen maturity (Table 1). Neither did the data suggest a trend toward a significant alteration in one of these molecular parameters. However, the analysis was based on a sample size of three, which is a limitation of the study. Hence, we cannot exclude the possibility that larger sample sizes may reveal differences in one of the molecular parameters. Together, loss of EGR-1 resulted in osteopenia without significantly affecting the molecular properties of bone. Such phenotypes have been reported before. For example, overexpression of the vitamin D receptor in mature osteoblasts resulted in increased BV/TV in mice [33], without affecting FTIRI parameters in cortical bone and a sole decrease in collagen maturity in trabecular bone [38], although changes in mineral crystal structure were detected by X-ray diffraction. Similar to EGR-1, vitamin D receptor is a convergence point between cellular signaling and gene-expression programs [39, 40].

A sex-specific function of EGR-1 was observed in trabecular bone of the femur. Male, but not female, Egr-1 −/− mice showed an osteopenic phenotype characterized by a statistically significant decrease in both BV/TV and BMD. Why female EGR-1-deficient mice are protected from bone loss is unclear. However, the study by Cenci et al. [32] showed that ovariectomy of Egr-1 −/− mice did not result in bone loss, suggesting that a minimal bone mass is guarded from further decrease. This may relate to our observations because a significant reduction of trabecular BV/TV was observed in wt female mice compared to wt male animals (Fig. 2a) and it is possible that the reduced female bone mass is shielded from further EGR-1-related reduction. In comparison to femur trabecular bone, femur cortical bone and rib bone BV/TV did not differ between wt male and female mice. Consistent with this observation, osteopenic characteristics were seen in both male and female rib bones of Egr-1 −/− mice (Fig. 5a). They were, however, statistically more distinct in female compared to male animals (Fig. 5a), a finding that needs further investigation.

Male and female bone properties were initially studied by micro-CT in femur (Figs. 2, 3, and 4) and tibia (see supplementary tables), both appendicular long bones. Measurements on femurs and tibias yielded very similar results. For example, in cortical bone the direction of change as well as the significance of all assessed parameters were identical between femur (Figs. 2, 4) and tibia (Supplementary Table 1). In contrast, rib bones showed a distinct alteration signature between Egr-1 −/− and wt mice (Fig. 5), suggesting a site-specific influence of EGR-1. Our study included ribs because they have been the subject of previous investigations, such as pharmacological studies of bisphosphonates [41, 42] and bone fracture healing [43, 44]. Very little, however, is known about rib bone biology, which makes it challenging to interpret our data regarding potential site-specific differences in bone biology that may explain the differential magnitude or specificity of EGR-1 activity between appendicular long bones and rib. Nevertheless, site-specific differences in bone cell activity have been reported before and are particularly well established for osteoclasts, although the mechanisms underlying these differences are largely unknown but may include variation in developmental origin, precursor populations, mechanical load, or hormonal activity [45]. Interestingly, early work by Wiktor-Jedrzejczak et al. suggested that the efficacy of rescue of the osteopetrotic phenotype of op/op mice with M-CSF was site-specific [46, 47]. Assuming loss of EGR-1 provokes a site-specific increase in M-CSF production [26], our findings are in line with these data. More recent work, however, indicates that site-specific differences in bone cell activity may stem from development-based gene-expression signatures, which are distinct for particular bones [48]. Further, studying human bones, Varanasi et al. [49] showed that site-specific alterations were due to differences in multiple signaling pathways rather than just cell numbers. Notably, a computational assessment of these signaling pathways using network-generating algorithms suggested, consistent with the present report, an involvement of EGR-1 in bone cell biology [49]. An important question is whether EGR-1 plays a functional role in the skeleton. We noted a significant reduction in the minimum moment of inertia in male cortical bone of the femur (Fig. 4d), which was suggestive of an alteration in mechanical properties. Subsequent testing revealed a decrease in several mechanical testing parameters during three-point bending (Fig. 4e–i). Hence, it is likely that the changes in cross-sectional geometry are the cause for the reduced resistance to three-point bending, but we also detected a small change in bone heterogeneity; thus, we cannot rule out the possibility that other factors may contribute to the observed change in femur mechanical properties. All together, this study demonstrated that both sex and skeletal site determine EGR-1 activity in vivo and that its site-specific action may influence the mechanical properties of bone.