Mammalian linear growth occurs at epiphyseal plates located at the ends of the principal long bones (humerus, radius/ulna, femur, and tibia/fibula). A typical long bone contains a growth plate at either end; however, these two physes significantly differ in their kinetics and resulting contribution to bone elongation. Most forelimb growth occurs at opposite ends of the limb at the proximal humerus and distal radius/ulna (i.e., the shoulder and wrist), while most hindlimb elongation occurs in the adjacent knee physes of the distal femur (DF) and proximal tibia (PT) at the center of the limb (Digby, 1915; Payton, 1932; Bisgard and Bisgard, 1935; Pritchett, 1991, 1992). These more active, faster-growing sites are characterized by higher rates of proliferation, cell turnover, and matrix mineralization compared to the less active proximal femur (PF) and distal tibia (DT; i.e., hip and ankle), which play only a minor role in bone lengthening (Kember, 1972; Wilsman et al., 1996a, 1996b). In addition, the faster-growing sites are also characterized by the greatest delays in epiphyseal closure across species (Dawson, 1925; Strong, 1925; Zoetis et al., 2003). These well-documented patterns are due largely to differences in cell kinetics inherent in each physis (Kember, 1978; Hunziker et al., 1987; Hunziker and Schenk, 1989; Hunziker, 1994; Wilsman et al., 1996a, 1996b), but the mechanisms underlying such striking within-bone variation have not been elucidated.
Growth plate regulation is complex, with resident chondrocytes capable of responding to a multitude of local and systemic factors, each having its own independent and potentially overlapping signaling cascade (i.e., van der Eerden et al., 2003; Nilsson et al., 2005a; Provot and Schipani, 2005). Such dynamics preclude easy resolution of the mechanism(s) that directly underlie differential growth plate activity, as it is unlikely that a single factor is alone responsible. However, the natural variation that exists between skeletal locations affords a unique opportunity to better understand growth plate regulation. For example, Reno et al. (2006) studied differential growth of the mouse metatarsal, a bone that only develops a true growth plate at one end, and identified unique patterns of chondrocyte behavior and protein expression specific to growth plate formation. By comparing chondrocyte populations from growth plates within and between different bones of the skeleton, it may be possible to isolate differences in protein expression that can be directly linked to variation in growth plate activity. While numerous possibilities exist, potent growth regulators such as growth hormone (GH) and insulin-like growth factor-I (IGF-I) emerge as key candidates because of their central role in normal growth. Gene misexpression studies have demonstrated that over- or underproduction of either of these molecules or their signaling components often results in disproportionate skeletal growth (Shea et al., 1990; Wolf et al., 1991; Sjögren et al., 2000, 2002; List et al., 2001), suggesting that they may partially underlie the variation in physeal growth rate naturally observed throughout the skeleton.
IGF-I is requisite in mediating the main GH response and is necessary for both embryonic and postnatal growth (Baker et al., 1993). Animals lacking IGF-I show a severe growth retardation that cannot be overcome by GH treatment, while GH-deficient animals show a less extreme growth reduction postnatally that can be reversed through IGF-I administration (Lupu et al., 2001; Isaksson, 2004). The major source of IGF-I is the liver from where it circulates with a complex network of facilitatory and inhibitory binding proteins (Le Roith, 2003). IGF-I can also be locally produced in cartilage (Isaksson et al., 1982; Nilsson et al., 1990; Reinecke et al., 2000), but high systemic concentrations of the hormone would likely diminish its ability to target skeletal growth plates differentially. However, as tissues can locally regulate their sensitivity to a circulating factor through up- or downregulation of receptors, the IGF-I receptor (IGF-IR) could be differentially expressed throughout the skeleton in such a way as to enable a disproportionate growth response.
The IGF-IR has been previously identified in cartilaginous growth plates via immunohistochemistry (Joseph et al., 1999; Visnapuu et al., 2001; Delatte et al., 2004; Eshet et al., 2004; Hoshi et al., 2004), in situ hybridization (Visnapuu et al., 2002; Cruickshank et al., 2005), Northern blot analysis (Smink et al., 2002), and RT-PCR (Oberbauer and Peng, 1995; Olney and Mougey, 1999), but prior research has not focused on comparisons within and between skeletal locations during the growth period. We therefore examined spatiotemporal expression patterns of the IGF-IR in the immature mouse hindlimb using immunohistochemistry. We tested the hypothesis that IGF-IR expression is directly related to growth plate activity. We predicted that the more active knee growth plates (DF and PT) would demonstrate increased receptor expression compared to the less active ends of the bones. Here, we present quantitative results of these analyses and discuss their implications for mammalian skeletal evolution and the processes of growth plate senescence and epiphyseal fusion.
MATERIALS AND METHODS
Animals and Tissues
All procedures were carried out in accordance with the Institutional Animal Care and Use Committee guidelines at the Northeastern Ohio Universities College of Medicine. C57BL/6 mice (n = 13) were housed in standard plastic caging with pine bedding, exposed to a 12-hr light/dark cycle, and provided with food (Formulab Rodent Diet, PMI Nutrition International) and water ad libitum. Animals were euthanized at 8, 16, and 28 days of age by CO2 asphyxiation and hindlimbs were immediately harvested. The upper age limit was chosen as 28 days to avoid complications introduced by puberty and the sexually dimorphic response known to accompany the GH-IGF axis (Jansson et al., 1985; Gatford et al., 1998).
Femora and tibiae were divided into proximal and distal segments, fixed in 10% neutral buffered formalin for 24 hr, and decalcified in 10% EDTA for 2–4 weeks. Samples were then rinsed in PBS, dehydrated in a graded series of ethanol, cleared in xylene, embedded in Paraplast Plus paraffin (Fisher), and cut into 6 μm sections using a rotary microtome (Leica RM2165).
Serial sections were stained using a rabbit polyclonal antibody against either the IGF-IRα (Santa Cruz Biotechnology, sc-712) or PCNA (Santa Cruz Biotechnology, sc-7907), at dilutions of 1:400 and 1:200, respectively. In both cases, antigen unmasking was accomplished by incubating slides in sodium citrate (pH 3.5) at 37°C for 30 min, followed by digestion with chondroitinase (Sigma) at 37°C for 10 min. Endogenous peroxidase was quenched with 3% H2O2 at room temperature (RT) for 10 min followed by incubation in goat serum (Santa Cruz Biotechnology) for 60 min at RT to block nonspecific binding. Incubation with the primary antibody occurred in a humidified chamber at RT overnight. Immunoreactivity was visualized using the ABC method (Santa Cruz Biotechnology) with DAB as the chromagen (Vector). Slides were counterstained with 0.1% methyl green (Sigma), dehydrated, and coverslipped with DPX (Sigma) prior to imaging. All quantifications were done on an Olympus BH-2 microscope interfaced to Bioquant Nova 5.0 image analysis software (BQIAC, Nashville, TN).
Chondrocytes positively stained for the IGF-IR were manually counted in 150 micron wide sample areas (three per section) spanning the entire height of the growth plate. A labeling index was calculated by dividing the number of positively stained cells by the total chondrocyte number so that relative comparisons could be made across ages and skeletal sites. PCNA immunoreactivity was quantified in the same manner, except cell counts were limited to positively stained nuclei in the columnar zone. Although this commonly used labeling method can be less specific for proliferating cells compared to more direct techniques such as incorporation of the thymidine analog BrdU (Muskhelishvili et al., 2003), we did not detect a significant difference between the methods in 28-day-old mice that had been injected with BrdU (data not shown), and others have successfully shown a strong correlation between the two techniques in bone and cartilage (Wildemann et al., 2003). PCNA was therefore preferred as a less invasive technique given the young ages and small size of the mice used here.
Growth Plate Morphometry
The height of the individual growth plate zones (reserve, columnar, and hypertrophic) was measured at each skeletal site. To account for natural size variation across the epiphysis, measurements were taken at three equidistant points spanning the width of the growth plate (10× objective) and the mean height of each zone was recorded. Total growth plate height was determined by the sum of the respective zones.
Growth Plate Morphometry
Total growth plate height declined with age in all sites examined (Fig. 1A). This decline was apparent in each of the growth plate zones, but the columnar zone showed the most marked reduction (Fig. 1B). There was also variation in overall growth plate size among the sites; the PF was the smallest in the 8-day-old mice, and the PF and DT both demonstrated the greatest reduction by 28 days.
PCNA expression was high in all sites at 8 days (Fig. 2). There was an age-associated decline in both relative and absolute staining in the PF and DT (Fig. 2B), but expression remained elevated in the DF and PT (knee), indicating continued proliferation and growth despite the observed decrease in epiphyseal height. It is interesting to note that PCNA expression was nearly equal among all sites in the 8-day-old mice. Although typically assumed to contribute little to adult limb length, these sites (PF and DT), traditionally viewed as less active, were found to be quite dynamic in early stages of ontogeny.
IGF-IR immunoreactivity was detected in all zones of the growth plate (Fig. 3A and C). There were very high levels of IGF-IR expressing cells in all sites in 8-day-old mice (Fig. 3), much like the proliferation results. Also similar to PCNA, there was an age-related loss in IGF-IR expression in the PF and DT. By 28 days, the IGF-IR labeling index had dropped over 2.5-fold from the levels measured at 8 days in these slower-growing physes (Fig. 3B), while maximum expression was maintained in the more active DF and PT. The relative decline in IGF-IR in the hip and ankle was most pronounced in the columnar zone (Fig. 3C), paralleling the disproportionate size reduction of this growth plate region (e.g., Fig. 1B).
IGF-IR and PCNA Comparison in Columnar Zone
There was a strong correlation between the absolute number of chondrocytes in the columnar zone expressing the IGF-IR and those expressing PCNA (Fig. 4; Pearson's r = 0.88; P < 0.001; n = 52 growth plates). Irrespective of age or site, growth plates with the highest number of chondrocytes expressing the IGF-IR also had the highest number of proliferating cells, suggesting that IGF-IR expression is associated with this increased growth plate activity.
Spatiotemporal Decline in IGF-IR
Our analysis revealed striking spatial and temporal variation in IGF-IR expression and growth plate activity in the immature mouse hindlimb. We found that the IGF-IR decreased with age only in the traditionally viewed “less active” sites of the hip and ankle (PF and DT), with the most notable decline occurring in the columnar zone, concomitant with its reduced proliferative activity. The DF and PT physes comprising the knee maintained maximal IGF-IR and PCNA expression at all ages examined here, indicating that high growth activity was sustained despite a noted reduction in overall growth plate size. This is unsurprising given the relatively large period of growth still remaining in mice of these young ages. Likewise, growth plate size is not a simple function of linear growth rate as Hunziker and Schenk (1989) aptly cautioned by means of their demonstration that high rates of growth could accompany growth plate reduction. This may also clarify why the PF was smaller in size compared to the other sites at 8 days and yet still displayed maximum levels of PCNA and IGF-IR expression.
The marked decline in IGF-IR and PCNA expression in the hips and ankles of 28-day-old mice is not inconsequential. According to published growth curves general for this inbred strain, a 4-week-old mouse has attained little more than half of its adult body weight and has a minimum of 8 additional weeks of growth remaining (Charles River Laboratories, 2006). Although body weight may not necessarily equate to skeletal size, skeletal maturity is often not considered to occur until at least 12 weeks of age in mice (Kilborn et al., 2002; Zoetis et al., 2003). Yet we have demonstrated, as have previous studies, that growth in the 28-day-old mice had nearly ceased in two major hindlimb growth plates, leaving the active knee physes alone to drive the remaining bone elongation. We have also shown that this sharp decline in physeal activity is strongly correlated with expression of the IGF-IR. This suggests that the IGF-IR is directly related to differential growth plate activity within and between growing bones of the skeleton.
Implications for Growth Plate Senescence
The age- and site-specific decrease in the IGF-IR and associated growth activity suggests that its expression may be linked with the normal processes of growth plate senescence (aging) and epiphyseal fusion. Although many rodents do not fully fuse their growth plates, they clearly experience an age-associated growth decline with a dramatic reduction in growth plate height. Roach et al. (2003) have identified changes in physeal structure that essentially prevent further bone elongation, and it is now understood that growth cessation is not a simple consequence of epiphyseal closure (Parfitt, 2002; Nilsson and Baron, 2004). The mechanisms underlying these complex changes are unclear, but estrogen has classically been viewed as the principal determinant of this process (Turner et al., 1994). More recent work, however, suggests that the effects of estrogen are not direct, but rather mediated through other local mechanisms within the growth plate itself (Weise et al., 2001; Nilsson and Baron, 2004, 2005; Rodd et al., 2004; Nilsson et al., 2005b; Schrier et al., 2006). Here, we propose that downregulation of the IGF-IR is among the factors driving senescent growth decline. Research involving the GH-IGF-I axis lends support to this idea. Testosterone has been shown to increase IGF-IR levels in the growth plate (Phillip et al., 2001). Eshet et al. (2004) found that inhibiting aromatase activity (and thus blocking the conversion of testosterone to estrogen) resulted in increased growth plate height and IGF-IR expression in mice, suggesting that estrogen may act locally to downregulate the IGF-IR. Interestingly, estrogen receptor knockout mice exhibit disproportionate skeletal growth that mimics the phenotype observed in some animals with disrupted GH-IGF axes (Vidal et al., 1999). Hoshi et al. (2004) reported that insufficient signaling of the IGF-IR through a deficiency in insulin receptor substrate-1 (IRS-1; the major second messenger of the IGF-IR) was linked to growth plate reduction and early epiphyseal closure, further suggesting that the IGF-IR signaling pathway plays an important role in maintaining an open and active growth plate. Furthermore, animals lacking GH or the GH receptor also have decreased epiphyseal height, likely due to the absence of GH and, as a result, IGF-I signaling (Lupu et al., 2001).
Regulation of IGF-IR
That sensitivity to a potent growth stimulator should be decreased in growth plates undergoing the process of senescence is not unexpected, but little is known about the precise regulation of the IGF-IR. In regard to local feedback, sex steroids can alter IGF-IR expression as noted above, and others have shown that GH and IGF-I both influence IGF-IR mRNA levels in vitro (Oberbauer and Peng, 1995). Additional in vitro experiments have demonstrated that the circulating hormone leptin (Maor et al., 2002), as well as steroids such as glucocorticoids (Smink et al., 2002), thyroid hormone (Ohlsson et al., 1992), and vitamin-D (Krohn et al., 2003), can also increase IGF-R expression in cartilage. From a genomic basis, Williams et al. (2005) recently identified two major components of the GH-IGF axis, the GH receptor and IGF binding protein-4, as candidate downstream genes regulated by HOXA13, a key transcription factor involved in embryonic patterning. While the IGF-IR was not specifically identified in their study, the fact that other critical components of its signaling pathway were upregulated by HOX suggests that the IGF-IR may also be under the direct control of a similar global patterning gene. Whole bone and isolated growth plate transplantation experiments have been invaluable in demonstrating that growth potential is an intrinsic property of the donor physis irrespective of host age or hormonal milieu (Felts, 1959; Chalmers, 1965; Harkness, 1974; Kline et al., 1990; Glickman et al., 2000), further suggesting that the local factors regulating growth plate activity (including the IGF-IR, as implicated by our results) are indeed the product of a genetic program established during early embryogenesis.
As a final point, we suggest that the IGF-IR is a potential target for skeletal evolution in mammals. By regulating the sensitivity of specific skeletal sites to a major growth factor, modification in IGF-IR expression at the level of the growth plate could be a local means of increasing or decreasing bone growth without altering systemic IGF-I levels. From an evolutionary perspective, differences in local epiphyseal growth have produced an amazing level of diversity in mammalian limb proportions. Humans, for instance, have longer hindlimbs and shorter forelimbs when compared to our closest living great ape relatives. Reno et al. (2005) demonstrated that the long hindlimbs of humans results in part from differential growth at the distal end of the femur. Based on the position of the nutrient foramen, they showed that humans have longer distal femora relative to size when compared to chimpanzees and gorillas. Indeed, the distal segment of the femur in humans was found to comprise a greater proportion of the bone's total length than that in apes, indicating that our femora are longer because of increased distal growth (Reno et al., 2005). Whether variation in the IGF system underlies these differences is unknown at this time, but our next step is to compare the spatial and temporal distribution of the IGF-IR in human and ape growth plates as tissues become available in order to investigate this compelling issue further.
The authors thank Drs. Walter Horton, John Stalvey, Phil Reno, and Rieko Yagi, as well as Ling Yang and Ashleigh Nugent, for useful comments and valuable discussion involving this project. Heather McEwen offered helpful research assistance and technical advice.