Altered IGF-I activity and accelerated bone elongation in growth plates precede excess weight gain in a mouse model of juvenile obesity

Abstract

Nearly one-third of children in the United States are overweight or obese by their preteens. Tall stature and accelerated bone elongation are characteristic features of childhood obesity, which cooccur with conditions such as limb bowing, slipped epiphyses, and fractures. Children with obesity paradoxically have normal circulating IGF-I, the major growth-stimulating hormone. Here, we describe and validate a mouse model of excess dietary fat to examine mechanisms of growth acceleration in obesity. We used in vivo multiphoton imaging and immunostaining to test the hypothesis that high-fat diet increases IGF-I activity and alters growth plate structure before the onset of obesity. We tracked bone and body growth in male and female C57BL/6 mice (n = 114) on high-fat (60% kcal fat) or control (10% kcal fat) diets from weaning (3 wk) to skeletal maturity (12 wk). Tibial and tail elongation rates increased after brief (1–2 wk) high-fat diet exposure without altering serum IGF-I. Femoral bone density and growth plate size were increased, but growth plates were disorganized in not-yet-obese high-fat diet mice. Multiphoton imaging revealed more IGF-I in the vasculature surrounding growth plates of high-fat diet mice and increased uptake when vascular levels peaked. High-fat diet growth plates had more activated IGF-I receptors and fewer inhibitory binding proteins, suggesting increased IGF-I bioavailability in growth plates. These results, which parallel pediatric growth patterns, highlight the fundamental role of diet in the earliest stages of developing obesity-related skeletal complications and validate the utility of the model for future studies aimed at determining mechanisms of diet-enhanced bone lengthening.

NEW & NOTEWORTHY This paper validates a mouse model of linear growth acceleration in juvenile obesity. We demonstrate that high-fat diet induces rapid increases in bone elongation rate that precede excess weight gain and parallel pediatric growth. By imaging IGF-I delivery to growth plates in vivo, we reveal novel diet-induced changes in IGF-I uptake and activity. These results are important for understanding the sequelae of musculoskeletal complications that accompany advanced bone age and obesity in children.

INTRODUCTION

Youth obesity rates have almost quadrupled since 1970 (1). Nearly one-third of children in the United States are either overweight or obese by age 10 (1, 2) and obesity is already prevalent in children as young as 2 (1). This point is important because high body mass index (BMI) during childhood is a strong predictor of future obesity and a suite of metabolic and musculoskeletal complications in adolescents and adults (3, 4). Children who gain excess weight and become progressively obese typically have higher rates of linear growth and skeletal maturation when compared with their nonobese peers (5–8). In fact, the relationship between obesity and accelerated linear growth is so characteristic that clinical guidance suggests a child with obesity that is not tall for their age should be evaluated for an underlying endocrine disorder (9). Although transiently taller than lean age-matched counterparts, children with obesity reach skeletal maturity and attain adult height sooner so they do not end up taller as adults (10). This short-term linear growth acceleration, which is already evident by preschool age (11, 12), often cooccurs with painful orthopedic conditions such as limb bowing, joint instability, increased fractures, slipped epiphyses, and early-onset osteoarthritis (13–17).

One of the major obstacles to preventing at least some of the skeletal complications of childhood obesity is that the mediators of obesity enhanced bone lengthening are not well understood. The presence of growth acceleration is often viewed as a biomarker of negative bone quality in overweight pediatric patients (18, 19). However, there is robust evidence that accelerated growth actually precedes the onset of obesity (18, 19), suggesting that there are physiological effects on the skeleton before a child’s weight exceeds normal parameters.

Bone lengthening occurs by the process of endochondral ossification in cartilaginous growth plates that are located at the ends of developing long bones (20). This process is driven by a complex interplay of signaling molecules that are both systemic and local to a heterogenous population of resident chondrocytes in the growth plate (21–23). Although there are hundreds of genetic variants known to control human height (24), the primary regulators are growth hormone (GH) and its mediator, insulin-like growth factor (IGF)-I (reviewed in (25)). IGF-I is indispensable for bone and body growth (26, 27). Surprisingly, children with obesity typically have low circulating levels of GH and normal IGF-I despite their accelerated growth (28–30). There is evidence that obesity might alter IGF-I bioavailability by reducing serum IGF binding proteins (IGFBPs), which normally inhibit growth by sequestering free IGF-I (29, 31–33). However, there are no data on local IGF-I uptake and bioactivity in growth plates during obesity.

We previously described a protocol for labeling, validating, and visualizing uptake of biologically active IGF-I in growth plates of live mice using in vivo multiphoton microscopy (34). Here we apply this method to quantitatively assess IGF-I delivery and uptake in growth plates using a high-fat diet to model the development of juvenile obesity in mice. We define “delivery” as the amount of IGF-I that reaches the metaphyseal bone vasculature, whereas “uptake” is the total amount that gets into the growth plate. We coupled this approach with postmortem protein assays by immunostaining to identify local biomarkers of IGF-I activity in growth plates. Our goals of the study were twofold: 1) characterize and validate a mouse model of diet-enhanced bone lengthening that mimics the pattern of enhanced statural growth during the development of juvenile obesity in humans and 2) identify a potential mechanistic role of IGF-I in the diet-induced growth response. We tested the hypothesis that a high-fat diet increases IGF-I activity and alters growth plate structure before the overt onset of obesity.

METHODS

Animals and Diets

All procedures were reviewed and approved by the Institutional Animal Care and Use Committee at Marshall University (protocols 527 and 558). Male and female C57BL/6 mice (n = 114 total) were obtained at 3 wk weaning age from a commercial vendor (Hilltop Lab Animals; n = 104) or from an in-house breeding colony (n = 10). With the exception of food intake studies (detailed below), mice were housed three per cage in conventional shoebox-style cages with cardboard huts, Enrich-n’Nest bedding (Cincinnati Lab Supply), and ad libitum access to food and water. Room temperature was 21°C with a standard 12-h light/dark cycle. Mice in food intake studies were singly caged under otherwise identical conditions. Experiments were performed during nine separate trials at different times of the year to account for potential seasonal variation.

At 3 wk age, mice were separated into size-matched groups based on body mass and randomly assigned to control or high-fat diets. All mice in the high-fat group (n = 57 total) received a standardized diet containing 60% energy from fat (TestDiet 58Y1, blue, 5.1 kcal/g). Most controls (n = 52 total) received a low-fat diet with 10% energy from fat (TestDiet 58Y2, yellow, 3.76 kcal/g). Although the control diet was only slightly different from standard rodent chow, both diets were semipurified (AIN-76A) to ensure that vitamin, protein, mineral, and fat components were the same, with only differing fat and carbohydrate proportions (control: 18% protein, 10.2% fat, 71.8% carbohydrate; high-fat: 18.1% protein, 61.6% fat, 20.3% carbohydrate). Controls used for multiphoton imaging experiments (n = 5 total) received a 13% energy from fat standard chow diet (LabDiet 5001, 4.09 kcal/g) because those data were collected before control diets were standardized.

Separate experiments were performed at three endpoints, when mice were euthanized by CO₂ asphyxiation for serum and tissue collection: 4 wk, 5 wk and 12 wk age (1, 2, and 9 wk of diet treatment). The 4- and 5-wk age points were chosen because we have previously found that these ages represent a time of rapid, environmentally sensitive growth in mice (35, 36). The 3- to 5-wk age period in mice is roughly comparable to human development between toddler and preteen ages (37). Some experiments continued to 12 wk age to assess long-term effects when mice are considered both skeletally mature (38) and obese from high-fat diet feeding (39). We did not perform experiments in males at the 4-wk age point due to lack of animal availability; however, our results showed that males and females followed the same patterns at the two other age points. All other groups utilized a mixed-sex sample.

To assess caloric intake, we measured food consumption in a subset of singly caged mice from the 5-wk age group (n = 8 per sex per diet as shown in Table 2) by measuring the difference between the amount offered the previous day and the amount remaining in the hopper and recovered from the cage floor the next morning following our published methods (40). The Enrich-n’Nest bedding is a rolled white paper bedding that enables accurate discrimination of color-coded food scraps from surrounding bedding. Measurements were recorded each day for 15 days and averaged for each mouse.

Sample Sizes and Statistical Analyses

Sample sizes (minimum n = 5–6 mice per variable unless otherwise stated) were determined a priori by estimating the effect size and data variability to yield a statistical power of 80% at α = 0.05. Most experimental trials were repeated twice to confirm results, although some data were only collected from mice during a single trial. Other missing data are the result of sample loss due to damage during dissection, processing, sectioning and/or staining, absence of fluorochrome label, or technical problems collecting samples and morphometric data. Missing cases were excluded from statistical testing on an analysis-by-analysis basis. Sample sizes for each analysis are detailed in Tables 1, 2 and 3 and Figure legends.

Table 1.

Morphometric data in 4-wk-old mice (1-wk diet treatment)

	Female
Parameter	n	Control	High fat	Cohen’s d	P value
Body mass at start, g	12	8.6 (1.3)	8.6 (1.2)	0.0
Body mass at endpoint, g	12	13.2 (1.4)	15.0 (1.1)	1.4	<0.001
BMI, kg/m2	12	2.6 (0.2)	2.8 (0.1)	1.2	0.005
Body length, mm	12	71.3 (1.9)	73.6 (2.3)	1.1	0.007
Tail elongation rate, mm/day	12	1.1 (0.2)	1.6 (0.3)	2.0	<0.001
Tibial elongation rate, µm/day	6	156 (15)	187 (8)	2.6	<0.001
Femur length, mm	6	11.6 (0.4)	11.9 (0.2)	0.9
Bone density mid-femur, mg/cm3	6	774 (28)	762 (11)	0.6

Values are mean (SD). Sample sizes indicate mice in each diet per variable. Cohen’s d, the standardized difference between the group means, provides a measure of effect size independent of the original measurement unit. A Cohen’s d of 1 (large effect size) indicates that the group means differ by 1 SD as described in methods. One-tailed P values (analyzed by independent samples t tests) are listed for significant pairwise comparisons. BMI, body mass index.

Table 2.

Morphometric and food intake data in 5-wk-old mice (2-wk diet treatment)

	Female					Male
Parameter	n	Control	High fat	Cohen’s d	P value	n	Control	High fat	Cohen’s d	P value
Body mass at start, g	14	8.8 (0.8)	8.8 (0.9)	0.0		14	9.2 (0.8)	9.2 (0.8)	0.0
Body mass at endpoint, g	14	16.5 (0.6)	17.5 (0.9)	1.3	0.001	14	20.5 (1.1)	21.2 (1.8)	0.5
BMI, kg/m2	14	2.9 (0.1)	2.9 (0.1)	0.0		14	3.2 (0.1)	3.2 (0.2)	0.0
Food consumption, g/day	8	3.3 (0.1)	2.7 (0.3)	2.7	<0.001	8	3.6 (0.2)	3.0 (0.3)	2.4	<0.001
Caloric intake, kcal/day	8	12.6 (0.3)	13.9 (1.3)	1.4	0.008	8	13.6 (0.7)	15.1 (1.8)	1.1	0.025
Rump skinfold, mm	6	1.1 (0.07)	1.3 (0.05)	3.3	<0.001	5/4	1.2 (0.04)	1.3 (0.04)	2.5	0.047
Body length, mm	14	75.1 (1.4)	77.1 (1.8)	1.2	0.002	14	80.4 (1.5)	81.9 (1.8)	0.9	0.013
Tail elongation rate, mm/day	14	0.81 (0.13)	0.89 (0.12)	0.6	0.049	14	0.85 (0.16)	1.02 (0.11)	1.2	0.002
Tibial elongation rate, µm/day	5/7	117 (7)	136 (9)	2.4	0.001	4/6	152 (6)	168 (14)	1.5	0.034
Femur length, mm	8	12.4 (0.2)	12.7 (0.2)	1.5	0.015	8	12.8 (0.2)	13.1 (0.3)	1.2	0.021
Bone density mid-femur, mg/cm3	5	776 (11)	808 (19)	2.1	0.006	6/5	805 (14)	831 (8)	2.3	0.003

Values are mean (SD). Sample sizes indicate mice in each diet per variable. Unequal sample sizes (listed as control/high-fat) were due to differences in data availability as described in methods. Variables with a significant sex-by-diet interaction (P < 0.05, assessed by two-way ANOVA) indicated in bold type. Cohen’s d is a standardized measure of effect size as detailed in Table 1. Sex-specific differences in high-fat diet group relative to control and their one-tailed P values (analyzed by independent samples t tests) are listed for significant pairwise comparisons. BMI, body mass index.

Table 3.

Morphometric data in 12-wk-old mice (9-wk diet treatment)

	Female					Male
Parameter	n	Control	High fat	Cohen’s d	P value	n	Control	High fat	Cohen’s d	P value
Body mass at start, g	6	8.6 (1.2)	8.5 (1.1)	0.0		6	9.6 (0.4)	9.4 (0.8)	0.0
Body mass at endpoint, g	6	20.9 (1.4)	25.0 (2.6)	2.0	0.004	6	24.6 (1.1)	35.0 (2.3)	5.8	<0.001
BMI, kg/m2	6	2.8 (0.2)	3.3 (0.3)	2.0	0.004	6	3.1 (0.1)	4.2 (0.3)	4.9	<0.001
Body length, mm	6	86.5 (0.5)	86.9 (1.1)	0.5		6	89.7 (1.2)	91.3 (1.2)	1.3	0.02
Tail length, mm	6	84.4 (1.5)	82.6 (0.6)	1.6	0.011	6	86.0 (1.6)	85.2 (1.9)	0.5
Tibial length at 3-wk start, mm	6	12.7 (0.5)	12.7 (0.3)	0.0		6	13.3 (0.1)	13.2 (0.4)	0.3
Tibial length at 8-wk midpoint, mm	6	16.0 (0.1)	16.1 (0.1)	1.0	0.026	6	16.5 (0.2)	16.7 (0.1)	1.3	0.032
Tibial length at 12-wk endpoint, mm	6	16.5 (0.1)	16.5 (0.2)	0.0		6	16.9 (0.1)	16.9 (0.1)	0.0
Femur length, mm	6	15.1 (0.1)	14.7 (0.2)	2.5	0.002	6	15.3 (0.3)	15.1 (0.3)	0.7
Bone density mid-femur, mg/cm3	6	1008 (32)	1008 (42)	0.0		6	1001 (18)	1010 (28)	0.4

Values are mean (SD). Sample sizes indicate mice in each diet per variable. Variables with a significant sex by diet interaction (P < 0.05, assessed by two-way ANOVA) are indicated in bold type. Cohen’s d is a standardized measure of effect size as detailed in Table 1. Sex-specific differences in high-fat diet group relative to control and their one-tailed P values (analyzed by independent samples t tests) are listed for significant pairwise comparisons. BMI, body mass index.

All image-based data were collected at least twice to ensure reproducibility. Either two individuals collected measurements separately and observer averages were used, or a single individual collected measurements on two separate days and averages were used. Statistical analyses were performed in SPSS 26.0 software with α = 0.05 as accepted significance. Residuals were examined for each variable by plotting all data points to confirm that outliers and/or skewness did not preclude parametric testing. Except where noted, analyses utilized one-tailed parametric tests because of the a priori hypothesis that growth would be increased in high-fat diet mice after treatment. Where applicable, data were first analyzed by two-way ANOVA to test for sex by diet interactions. Each sex was subsequently analyzed separately by independent samples t tests to more clearly assess diet effects. Tail length and body mass in the 12-wk sample were assessed by repeated measures ANOVA to test for diet effects over time. Serum cytokines were assessed in R statistical software (version 4.0.3) by a two-tailed nonparametric Mann–Whitney test corrected by the Holm method to account for small sample sizes and variability in cytokines, assuming no a priori differences between diets.

Since data were collected using different techniques, Cohen’s d was calculated to provide an index of effect size independent of the original measurement unit (Tables 1, 2 and 3). Cohen’s d is the standardized difference between group means that can accompany a P value to indicate the size of an effect (41, 42). For example, a Cohen’s d of 1 indicates that the group means differ by 1 standard deviation and a Cohen’s d of 2 indicates that the group means differ by 2 standard deviations. Cohen (1988) operationally defined 0.2, 0.5, and 0.8 as small, medium, and large effect sizes, respectively (41).

Body Size and Morphometry

Body mass and tail length were recorded weekly as detailed in the Supplemental Methods. Tibial length was measured from X-rays at the start (3 wk age), midpoint (8 wk age), and end point (12 wk age) in the 12-wk cohort (n = 24 mice total) following our published methods (43) and detailed in Supplemental Methods. Femora collected at each age end point were dissected, cleaned, and measured to obtain bone length and midshaft cortical density (see Supplemental Methods).

To obtain a relative measurement of subcutaneous fat in the 5-wk sample, we measured rump skinfold thickness immediately following euthanasia on a shaved region of the dorsal skin between the hindlimbs using a Starrett dial micrometer (model 1010). We also calculated body mass index (BMI), which has been shown to positively correlate with carcass fat in rodents (44, 45). We used morphometric data recorded at euthanasia to calculate BMI = body mass (kg)/nose-anus body length² (m²). Since there is no standardized definition of obesity in mice, we used Centers for Disease Control and Prevention (CDC) criteria for defining overweight and obesity status in children by calculating age- and sex-specific BMI percentiles (46). Overweight is defined as BMI in the 85th to less than 95th percentile compared with the age- and sex-matched cohort (diets combined), whereas obese is defined at the 95th percentile or greater (46). Although other approaches have defined obesity in adult mice using body mass at +3 standard deviations of the controls (47), we used BMI percentiles that consider mass for a given length to more closely match criteria used to define obesity in growing children.

We acknowledge that skinfold, body mass, and BMI only provide an indicator of obesity compared with more comprehensive body composition analyses that incorporate DEXA (dual-energy X-ray absorptiometry), fat pads, and adipocyte morphology. Body weight and length are relevant because they are comparable to data obtained during pediatric well-visit checks.

Serum Cytokine Assay

We measured serum cytokines in a subset of 5-wk-old males (N = 6 per diet) using a mouse obesity ELISA strip with chemiluminescence detection (Signosis, Inc., Santa Clara CA, EA-1721) to compare relative expression levels of 8 different proteins that are known to be altered in obesity-related disorders: leptin, TNFα, IGF-I (total), IL-6, VEGF, IL-1α, IL-1β, and MCP-1. This ELISA strip is designed for simultaneous comparison of 8 cytokines in 12 different samples. We chose the 5-wk age point because it is within a critical growth period when the skeleton is most responsive to environmental stimuli (35), and so we wanted to test whether we could detect serum biomarkers of inflammation linked to diet. We did not obtain sufficient samples to perform analyses on the 4- or 5-wk-old females.

Blood was collected by terminal cardiac puncture following published methods (48). We used 40-µL total serum (stored at −80°C until use) per mouse and performed the assay following the manufacturer’s instructions. Luminescence intensities were read immediately on a Spectramax i3x plate reader set to an integration time of 1 s and no filters. Expression levels of the cytokines were compared between individual mice. Standards were not included in the assay, so the results were semiquantitative in that relative cytokine levels could only be compared among individuals assayed on a single plate as others have noted (49). Due to the limited sample sizes and data variability, cytokine data were analyzed by nonparametric statistical tests as detailed above.

In Vivo Multiphoton Microscopy

A separate mixed-sex sample of 5-wk-old mice (n = 5 per diet) was used to quantify diet effects on IGF-I uptake in growth plates after 2 wk high-fat diet exposure. We followed our published protocol for labeling, validating, and visualizing uptake of biologically active IGF-I in growth plates in vivo (34) as detailed in the Supplemental Methods.

Mice were given a single 100-µL intraperitoneal injection of fluorescently labeled IGF-I (∼90 or ∼6 µg/gram body wt in a 15–16 g mouse). This dose is within a range that stimulates bone growth in mice (50) and could be visualized in the growth plate in vivo. Z-series images from superficial to deep were collected immediately before the injection and at 15-min intervals spanning 90 min after the injection using simultaneous collection channels for collagen in the perichondrium and fluorescently labeled IGF-I (34). Growth plate location was verified at each time point by matching morphology of the collagen in the z-series.

All image processing was done in ImageJ software to quantify IGF-I fluorescence intensity in 150 µm × 400 µm regions in the growth plate and metaphyseal vasculature at a standardized depth of 25 µm below the deepest edge of the perichondrium following our prior methods (51). Due to absolute and relative differences in vascular fluorescence between diets (see Fig. 3B), we were unable to utilize growth plate/vascular ratios to compare growth plate levels between animals as we have done previously (51, 52) since there would be a shifting denominator. To quantitatively assess how diet alters relative IGF-I uptake in growth plates, we calculated a time-dependent change in fluorescence intensity in the growth plate during the time period spanning 45–75 min after the IGF-I injection, which is when vascular levels peaked in both groups. This formula [(fluorescence₇₅ – fluorescence₄₅)/fluorescence₄₅] × 100 provided a relative measure of change in time-dependent growth plate uptake when vascular levels were maximum and enabled us to better discriminate diet effects on systemic IGF-I uptake in the growth plate.

Histomorphometry and Immunostaining

Mice were given a single intraperitoneal injection of oxytetracycline (OTC; 7.5 mg/kg, Norbrook 200 mg/mL) at either the start (4-wk age group) or midpoint of the study (5-wk age group) to measure tibial elongation rate. We did not measure elongation rate in the skeletally mature 12-wk age group. The proximal tibial growth plate was selected because of its relatively flat contour and uniform growth rate (53). The tibia was cleaned of soft tissue and bisected longitudinally, perpendicular to the growth plate, by making a single midline cut through the anterior tibial crest using a number 11 scalpel blade. Samples were rendered unusable if the tibial cut was not through the midline and perpendicular to the growth plate. One half of the bisected tibia was placed in a customized holder on a glass slide and cover-slipped with glycerol in PBS. The other tibial half was fixed for 24 h at 4°C in 4% paraformaldehyde (Biotium, 22023) for histological processing. OTC bones were stored in the dark at 4°C and imaged to measure tibial elongation rate (detailed in Supplemental Methods).

The fixed tibial halves were decalcified for 1 wk in 19% EDTA (pH 7.4) at 4°C, processed, embedded in paraffin, and cut into 6-µm sections using a rotary microtome. Bisected tibiae were positioned with the cut edge of the growth plate fully against the bottom of the mold to ensure that microtome sections were made perpendicular to the growth plate. We followed our published immunostaining methods (54) using antibodies against the phosphorylated IGF-I receptor (pIGF-IR, 1:100; Abcam, Cat. No. ab38389) and IGF binding protein-4 (IGFBP-4, 1:400; Millipore, 06–109) in separate sections (additional details in Supplemental Methods). The phosphorylated IGF-I receptor was used to assess IGF-I bioactivity because it is a well-established biomarker of IGF-I signal activation (55–57). IGFBP-4 was selected because it functions primarily as an inhibitor of IGF-I actions (58, 59). IGFBP-4 is the most abundant binding protein in all cell types from bone (60) and is a well-established inhibitor of IGF-stimulated proliferation of bone and cartilage in vitro (58, 61, 62) and in vivo (63, 64). Growth plates were analyzed for morphology and protein expression following our published methods (54) (described in Supplemental Material).

RESULTS

Body Size and Morphometry

Morphometric data for each age, sex, and diet are detailed in Tables 1, 2 and 3. Mice on a high-fat diet had reduced food consumption, but their caloric intake was greater due to the higher fat content of the diet (Table 2). There was sexual dimorphism in most measurements; however, males and females exhibited similar patterns in response to diet in nearly all of the variables examined. The only sex by diet interactions that we detected were related to body mass and fat deposition. Females had relatively more subcutaneous fat, assessed by skinfold thickness, when compared with their male counterparts on a high-fat diet at 5 wk age (two-way ANOVA F = 18.13, P = 0.001; Table 2). Males on a high-fat diet had a relatively higher body mass (F = 15.99, P = 0.001) and BMI (F = 9.55, P = 0.006) compared with females at 12 wk age (Table 3). Males and females otherwise exhibited parallel changes in all other variables.

Diet effects on both tail and tibial elongation rates were evident within the first week of the experiments (Table 1). Tibial elongation rate (Fig. 1A) was over 2.5 standard deviations greater (Cohen’s d = 2.6) in 4-wk-old mice after only 1 wk on a high-fat diet (>30 µm/day increase in elongation rate; t = 4.42, P < 0.001), and tail elongation rate was 2 standard deviations greater (0.5 mm/day increase in growth rate; Table 1). By 5 wk age (Table 2), these differences in bone elongation rates translated to 0.3 mm increase in actual femoral length (>2% of total bone length; Fig. 1B; t = 2.42, P = 0.03) and nearly 1.5 mm more tail growth (nearly 20% of total tail length) after 2 wk on the high-fat diet (Fig. 1C; t = 2.86, P = 0.007). Midshaft femoral bone density was over 2 standard deviations higher at 5 wk age (Table 2).

Figure 1.

Diet effects on bone elongation rate and length. Boxplots show over a 2.5 standard deviation increase (Cohen’s d = 2.6) in tibial elongation rate (>30 µm/day increase) in 4-wk-old female mice after 1 wk on a high-fat diet (A). By 5-wk age (2 wk on diet), these differences in bone elongation rate translated to 0.3 mm increase in actual femoral length (>2% of total bone length; B) and nearly 1.5 mm more tail growth (nearly 20% of total tail length; C). Significant (P < 0.05) by independent samples t tests as indicated in each graph. The same subset of 5-wk-old female mice is plotted in B and C. Sample sizes are shown. Boxes show interquartile range (middle 50%), horizontal line denotes the median, and whiskers denote minimum and maximum values excluding outliers. All individual data points are shown. N indicates number of mice per group.

There were significant increases in nose-to-anus body length in the high-fat diet group at both 4- and 5-wk age points as detailed in Tables 1 and 2. Together with the tail data, these results indicate that the accelerated skeletal growth occurred systemically and was not limited to weight bearing bones. Most of these skeletal differences observed at the two early age points were attenuated or even reversed in the 12-wk skeletally mature mice, as demonstrated by decreased tail and femur lengths and similar bone density (Table 3). All cohorts had a similar diet-induced growth response with no overt differences due to seasonality.

Growth Patterns

Supplemental Fig. S1 (all Supplemental material is available at https://doi.org/10.6084/m9.figshare.16850215) shows changes in body mass and tail length over time in the male 12-wk age sample. There was a significant effect of diet on body mass (repeated measures ANOVA F = 71.3, P < 0.001). Post hoc testing done on the individual time points confirmed a significant increase in mass in the high-fat diet group beginning at 5 wk age and persisting through the end point (Supplemental Fig. S1). Although repeated measures ANOVA did reveal a significant diet effect on tail length (F = 6.06, P < 0.017), the pattern was distinct from the steady increase in body mass. Tail length in the high-fat diet group rapidly increased within the first week and remained longer than that of controls for the next 3 wk (4- to 7-wk age). Post hoc testing revealed a significant difference in absolute tail length at 6 wk age as shown in Supplemental Fig. S2. For clarification, the change in tail length, which accounts for differences in starting tail length between animals shown in Fig. 1C, is significantly greater in high-fat diet mice by 4 wk age indicating that their tails had already grown more than controls after only 1 wk on the diet. Absolute tail length in this subset of mice was not statistically different until 6 wk age because of the variation in their tail lengths at the start (Supplemental Fig. S1). Tail lengthening reached a plateau in the high-fat diet group around 7 wk age, whereas it continued in controls so that the transient increase in high-fat diet tail length was attenuated by the study end point (Supplemental Fig. S1). Females exhibited the same patterns, with a less-pronounced increase in body mass on a high-fat diet (Table 3).

Tibial length, measured at 3-, 8-, and 12-wk age in the same cohort of mice, matched the results observed in the tail. Tibial length did not differ between diets at the start or end of the study, but was significantly longer at the midpoint (Table 3) when tails were similarly lengthened in the high-fat diet group.

Body mass and BMI were both greater in the high-fat diet mice at 4 wk (1-wk on diet) and 12 wk age (9 wk on diet) as shown in Tables 1 and 3. Body mass was only slightly elevated in both sexes at the 5-wk age point (reaching statistical significance in the females), but BMI did not differ in either sex at 5-wk age (Table 2). Since BMI is calculated as a mass-to-length ratio (kg/m²), it is relevant that the most pronounced increase in body length in high-fat diet mice also occurred at the 5-wk age point (Table 2), potentially offsetting their increased body mass. Another relevant note is that just over half of the mice in the 5-wk age sample were singly caged for food intake studies, which may have impacted weight gain even though it did not affect bone elongation rate.

We used age- and sex-specific BMI percentiles to classify mice into overweight and obese categories based on CDC criteria for children as justified in methods. Supplemental Fig. S2 shows a scatterplot of body mass (g) versus nose-to-anus body length (mm) by age for all mice in the study. Mice classified as overweight (85th to less than 95th BMI percentile) or obese (greater than 95th BMI percentile) are noted in each graph. Body mass generally increased in proportion to length at all ages with substantial overlap between diets. All of the overweight and obese mice at the 4- and 12-wk age points were from the high-fat diet group as expected. However, over half of the overweight and obese mice in the 5-wk age group were controls (Supplemental Fig. S2). The increased body length of the high-fat diet mice appeared to offset their slightly increased mass. One surprising result was the low number of individuals marked overweight or obese in the 12-wk age group when body mass was significantly greater in the high-fat diet groups (see Supplemental Fig. S1 and Table 3). Since we did not have a reference population for BMI, a limitation is that the calculation is based on age- and sex-specific BMI percentiles for each cohort in our sample. By definition, this results in flagging only those mice with the highest BMI values that were greater than 85% of the age- and sex-matched sample, regardless of actual body mass differences. The main finding is that there were systemic increases in skeletal growth and subcutaneous fat in high-fat diet mice when their body mass and BMI were largely similar to controls. The growth acceleration preceded excess weight gain.

Serum Obesity Cytokines

We next wanted to determine whether there were changes in serum cytokines that occurred with the rapid increase in skeletal growth in high-fat diet mice. We analyzed relative levels of eight obesity-related cytokines in 5-wk-old male mice and found significant changes in TNFα, IL-6, and VEGF (Fig. 2). Although VEGF was significantly elevated in the high-fat diet group (P = 0.03 by two-tailed Mann–Whitney corrected by Holm method), there was unexpectedly a 1.9- and 1.5-fold decrease in TNFα (P = 0.017) and IL-6 (P = 0.049), respectively (Fig. 2). There were no differences in total serum IGF-I levels. The assay did not have a binding protein disassociation step, and so we were not able to discriminate free versus IGFBP-bound fractions. There were out-of-range outliers in the control group for both the IL-1β and MCP-1 assays (N = 1 each) that are not shown on Fig. 2 but were included in statistical testing. Samples were within range in repeated testing, but we could only compare individuals tested on the same plate since we did not have standards. All individual data points except for the two noted outliers are shown. Data were analyzed by a nonparametric approach because of the noted differences in variances for IL-1β and MCP-1, as well as leptin and IGF-I. In most cases, the control group had the larger coefficient of variation.

Figure 2.

Relative levels of obesity-related cytokines in 5-wk-old male mice. Although most inflammatory markers were similar between the diets, VEGF was significantly increased in the high-fat diet group, whereas TNFα and IL-6 were decreased (two-tailed Mann–Whitney corrected by Holm method). Total serum IGF-I did not differ despite accelerated growth in the high-fat diet mice. Boxes show interquartile range (middle 50%), horizontal line denotes the median, and whiskers denote minimum and maximum values excluding outliers. There were out-of-range outliers in the control group for both the IL-1β and MCP-1 assays (n = 1 each) that are not shown on the graph but were included in statistical testing. The outliers were omitted on the graph for visual representation only. The samples were within range in repeated testing, but we could only compare individuals tested on the same plate since we did not have standards. All individual data points (mice) except for the two noted outliers are shown. N indicates number of mice per group.

IGF-I Delivery to Growth Plates

Total serum IGF-I did not differ between diets in accordance with previous literature, and so we used in vivo multiphoton imaging to determine whether there were diet-related differences in IGF-I delivery (amount that reaches metaphyseal vasculature) and uptake (amount that gets into the growth plate) in the proximal tibia at 5 wk. Using biologically active, fluorescently labeled IGF-I, we quantified delivery and time-dependent uptake by measuring relative fluorescence intensity in defined regions of the growth plate and metaphyseal vasculature at standardized depths and time points. At peak vascular fluorescence, which occurred ∼60 min after the injection in both groups and matches our prior studies (34), there was absolutely more IGF-I in the metaphyseal vasculature of the high-fat diet group (Fig. 3, A and B), as shown by increased fluorescence intensity in Fig. 3A. As a proportion of the total fluorescence (growth plate + metaphyseal vasculature), there was relatively more in the vasculature of the high-fat diet mice at the 60-min time point (t = 1.87, P = 0.049).

Figure 3.

High-fat diet increases IGF-I delivery and uptake in proximal tibial growth plates of live 5-wk-old female mice. Multiphoton images (A) taken at a standardized depth (25 µm below deepest edge of perichondrium) 60 min after injection of biologically active, fluorescently labeled IGF-I show IGF-I accumulation in the growth plate (between arrowheads) and metaphyseal vasculature (bottom of images). IGF-I appears as the white fluorescence. B: at 60-min peak vascular fluorescence, the stacked bar graphs show that there was absolutely more IGF-I in the metaphyseal vasculature of the high-fat diet group, as shown by increased white fluorescence intensity in A. As a proportion of the total fluorescence (growth plate + metaphyseal vasculature), there was relatively more in the vasculature of the high-fat diet mice at the 60-min time point independent samples (t = 1.87, P = 0.049). C: the time-dependent increase in the growth plate during this peak, measured as relative change in growth plate fluorescence between 45 and 75 min after the injection (see methods for details), increased 10-fold more in the high-fat diet group compared with the controls, in which levels remained steady or declined. Although these results were biologically distinct and repeatable in the high-fat diet mice (C), they were not statistically significant due to an outlier in the control group. See discussion. Boxes show interquartile range (middle 50%), horizontal line denotes the median, and whiskers denote minimum and maximum values excluding outliers. All individual data points (mice) are shown. Stacked bar graphs in B show the mean fluorescence in each zone for visualization purposes.

When we examined time-dependent growth plate uptake during this peak (Fig. 3C), there was over a 10-fold increase in the high-fat diet group compared with the controls when measured as a relative time-dependent change in growth plate fluorescence between the 45- and 75-min time points. Growth plate intensities continued to increase during this time in high-fat diet mice, whereas they were steady or declining in controls, consistent with our prior control studies (34). Although these unique results were repeatable in the high-fat diet mice (Fig. 3C), they were not statistically significant due to an outlier in the control group (t = 1.60, P = 0.074). If this outlier is not included in statistical testing, the results do become statistically significant (t = 2.94, P = 0.011). However, without scientific justification for excluding the outlier, we cannot report this as a statistically significant finding. However, we believe these differences in the high-fat diet mice are biologically important and would likely reach statistical significance with more robust sampling. The most relevant finding is that there was increased IGF-I delivery to growth plates of the high-fat diet mice without differences in circulating IGF-I levels.

IGF-I Bioactivity in Growth Plates

Since our in vivo experiments suggested there was increased delivery of injected (exogenous) IGF-I to growth plates of high-fat diet mice, we sought to determine whether there was a change in endogenous IGF-I bioactivity. We analyzed postmortem proximal tibial growth plates in a separate sample of mice (not injected with IGF-I) for expression of the phosphorylated IGF-I receptor (pIGF-IR). Figure 4, A and B, shows a significant increase in the labeling index for the pIGF-IR in the high-fat diet group. Although an average of 65% of chondrocytes were positively stained for the pIGF-IR in controls, the labeling index exceeded 80% in high-fat diet mice (t = 2.37, P = 0.020), indicating they had relatively more activated IGF-I receptors consistent with their enhanced tibial elongation rate (Fig. 4A). Total IGF-I receptor number did not differ in a small sample (n = 4; t = 2.04, P = 0.086), suggesting the difference in activated receptors was not due to total receptor number.

Figure 4.

High-fat diet alters IGF-I signaling in growth plates. Boxplots show that 5-wk-old male mice on a high-fat diet have increased expression of the phosphorylated IGF-I receptor, pIGF-IR (A) but decreased IGFBP-4 (C), as indicated by increased brown staining in B and decreased staining in D in the high-fat diet group. Growth plate, which was larger in high-fat diet mice (see Fig. 5), is between the black arrowheads on the left side of each image. Positively stained cells, defined by the brown DAB color, were manually counted in a standardized region in the proliferative zone. Negatively stained cells (yellow arrowheads in B and D) were defined and manually counted by the absence of DAB coloration and a methyl-green stained nucleus. Data are shown as a labeling index, which represents the percentage of cells positively stained relative to total in the region of interest. Boxes show interquartile range (middle 50%), horizontal line denotes the median, and whiskers denote minimum and maximum values excluding outliers. All individual data points (mice) are shown. DAB, diaminobenzidine. Sample size (N) indicates number of mice and P value obtained by independent samples t test.

To test whether IGF binding proteins might play a role in the differences in IGF-I uptake and bioactivity, we next examined relative expression of IGFBP-4, a classic IGF-I inhibitor. Figure 4, C and D shows a significant decrease in the IGFBP-4 labeling index in high-fat diet mice (74% of chondrocytes) when compared with controls (91% of chondrocytes), suggesting that the high-fat diet mice had fewer inhibitory binding proteins in the growth plate (t = 5.88, P < 0.001). These results might help explain our somewhat paradoxical IGF-I uptake data (absolutely more in control growth plates, Fig. 3B), as it is possible that more IGF-I was actually tethered to IGFBPs rather than binding to and activating the IGF-I receptor (see discussion). The most relevant finding is that a high-fat diet increased the relative number of activated IGF-I receptors and decreased inhibitory binding proteins (IGFBP-4) in growth plates.

Growth Plate Morphology

Total height of the growth plate was increased in high-fat diet mice, with significant increases in proliferative (t = 4.20, P < 0.001) and hypertrophic zones (t = 6.08, P < 0.001), and no difference in the reserve zone (Fig. 5A). Although the growth plates were larger in the high-fat diet mice, there were fewer cells in the proliferative zone (Fig. 5B; t = 2.29, P = 0.021) and chondrocyte columns were less organized than those of controls as measured by their orientation relative to the longitudinal axis of the bone (Fig. 5C; t = 6.85, P < 0.001). Columns in the control mice aligned within ∼2° of the long axis of the bone, whereas chondrocyte columns in the high-fat diet group deviated nearly 8° from the longitudinal axis (Fig. 5C), giving the growth plates a somewhat disorganized appearance (see Figs. 4, B and D). In addition to expansion of the hypertrophic zone (Fig. 5A), hypertrophic chondrocytes were also larger in the high-fat diet mice as measured by cell areas (Fig. 5D; t = 4.82, P < 0.001). Although we only had sufficient sample sizes to report data at the 5-wk age point, we did observe similar growth plate changes as early as 4 wk age after only 1 wk on a high-fat diet. The main finding is that a high-fat diet increased growth plate height, hypertrophic cell area, and columnar disorganization. Coupling morphometric data with our in vivo uptake and ex vivo IGF-I bioactivity studies, these results together suggest that a high-fat diet dysregulates IGF-I signaling and accelerates bone elongation in growth plates of preobese juvenile mice.

Figure 5.

Growth plates are larger but less organized in 5-wk-old male mice on a high-fat diet. Stacked bar graphs (A) show average height of the individual growth plate zones in control and high-fat diets. Proliferative and hypertrophic zones were significantly enlarged in the high-fat diet group (P < 0.001 by independent samples t test as shown in graph). Boxplots show that mice on a high-fat diet had fewer chondrocytes in the proliferative zone (B) with less organized columns (C). Columns aligned within 2° of the longitudinal axis of the bone in controls but deviated nearly 8° in high-fat diet mice, indicating a disruption in columnar organization. In addition to expansion of the hypertrophic zone in A, hypertrophic chondrocytes were enlarged in the high-fat diet mice as measured by cell areas shown in D. Boxes show interquartile range (middle 50%), horizontal line denotes the median, and whiskers denote minimum and maximum values excluding outliers. All individual data points are shown in boxplots. Stacked bar graphs in A show the mean of each growth plate zone for visual representation. N indicates number of mice in each group.

DISCUSSION

Our goals were to 1) characterize and validate a mouse model of diet-enhanced bone lengthening that mimics the pattern of enhanced statural growth during the development of juvenile obesity in humans and 2) assess a potential mechanistic role of IGF-I in driving this process. We tested the hypothesis that a high-fat diet increases IGF-I activity and alters growth plate structure before the overt onset of obesity. Key findings that support this hypothesis and suggest local IGF-I signaling is dysregulated by a high-fat diet include increased IGF-I delivery and receptor activation in growth plates, decreased inhibitory binding proteins (IGFBP-4), increased bone density, growth plate height, and hypertrophic cell area, and decreased cell density and columnar organization in faster-growing high-fat diet mice with similar body mass and BMI. These highly reproducible changes, which parallel patterns of linear growth acceleration in children as discussed below, were evident in the first 2 wk on a high-fat diet during a critical environmentally sensitive window of heightened growth before mice became overtly obese.

Our study is not the first to document a rapid change in cell kinetics after only a brief period of high-fat diet exposure. Mosser et al. (65) demonstrated increased pancreatic β-cell proliferation in 8-wk-old mice after only 3 days on a high-fat diet before insulin resistance was apparent, suggesting that a high-fat diet might induce almost immediate cellular-level changes before characteristic features of obesity and diabetes are detectable. We have previously reported that the postweaning period represents a critical environmentally sensitive window of postnatal growth in mice (35). In fact, most of the changes we saw in total tail lengthening actually occurred during the first week of the experiments (3–4 wk age) well before mice became obese. Here, we document and describe how rapid changes in bone elongation rate, growth plate structure, and IGF-I bioactivity precede excess weight gain in juvenile mice on a high-fat diet during a growth period comparable to humans between toddler and preteen ages (37, 38).

Systemic Increase in Bone Elongation Rate Precedes Excess Weight Gain

Two of the main findings of our study are that 1) the increase in bone elongation rate is systemic (it occurs in both weight-bearing hindlimb and nonweight-bearing tail) and 2) accelerated bone elongation rate occurs before mice exhibit disproportionate weight gain. Although the biomechanical stress of excess weight is thought to play a role in the physiology and pathophysiology of bone formation in obesity (66–68), our results suggest that weight-bearing per se is not the primary driver of growth acceleration and bone structural changes in these young mice. We show that changes in elongation rates and lengths of the tibia (weight-bearing) and tail (nonweight-bearing), as well as body length, occur within the first 2 wk on a high-fat diet. Our results are consistent with other studies that have shown increased bone and body lengths in juvenile mice on a high-fat diet (69, 70). Neither body mass nor BMI were substantially greater at the age points when we saw the most notable changes in bone lengths (see Table 2), suggesting that mass alone was unlikely to drive these effects through excessive mechanical loading.

It is unclear why we saw a more pronounced increase in body mass in the 4-wk age group since the body mass differences were only moderate at 5 wk. Although we did not detect overt seasonal effects, it is possible that the variation in body mass could be due to seasonality since the 4- and 5-wk studies were conducted during different times of the year. Our studies were not specifically focused on detecting seasonal effects, but others have documented seasonal variation in skeletal and body composition in C57BL/6 mice (71) and should be considered going forward. We also cannot discount the potential impact of single-housing on weight gain in some of the 5-wk mice (see Caveats and Limitations); however, nearly half of the mice in the 5-wk sample were not singly caged and exhibited the same trends. Even with similarities in overall body mass in the 5-wk age group, there were significant increases in skinfold thickness (subcutaneous fat), indicating the mice did differ in body composition. Additional body composition data, including fat pads and marrow adipose content, would provide a more robust assessment of obesity as others have done (72–74).

Interestingly, bone density was also increased in the 5-wk-old high-fat diet mice but did not differ at the earliest (4 wk) or latest (12 wk) age points. The impact of obesity on bone density is not fully understood (67, 75). Kondiboyna et al. (76) found that adolescents with obesity had higher apparent bone density compared with normal weight children, but actually had lower relative bone density when normalized to body mass. Obesity tends to increase bone density in adults, but with a paradoxical increase in fracture risk (75, 77, 78). The relevance of our bone density data is that a high-fat diet appears to induce a suite of rapid morphological changes that extend beyond bone lengthening and occur before the onset of obesity. More comprehensive studies including trabecular histomorphometry are needed to better understand how the faster-elongating growth plates directly alter bone density and architecture at the chondro-osseous junction.

Enhanced IGF-I Bioactivity but Disorganized Growth Plates

Our results are consistent with pediatric literature that indicate growth acceleration occurs in overweight children without a significant increase in circulating IGF-I (28–30). Our in vivo multiphoton imaging experiments suggest that the accelerated growth might occur through a site-specific increase in IGF-I delivery to skeletal growth plates. We detected absolutely more IGF-I in the metaphyseal vasculature of the high-fat diet mice (Fig. 3B). The time-dependent increase in IGF-I uptake in high-fat diet growth plates at the vascular peak (Fig. 3C) is also a novel finding. In our prior experiments (34), we typically saw steady to declining IGF-I levels in the growth plate when vascular levels peaked. The increase in the high-fat diet group is consistent with our findings of increased IGF-I receptor activation (Fig. 4A) and enhanced elongation rate (Fig. 1), suggesting that not only was more IGF-I getting into the high-fat diet growth plates, but that it appeared to be activating the IGF-I receptor and initiating growth-stimulating pathways. The latter point is relevant because we cannot rule out binding protein interference in our control experiments as discussed further below (see Caveats and Limitations).

The increased growth plate height in the high-fat diet mice is also consistent with increased bone elongation rates, which largely depend on components of the IGF-I signaling axis (reviewed in Ref. 25). However, there were distinct structural differences in the high-fat diet group that indicated these were not simply normal growth plates elongating at a faster rate. For instance, proliferative zone expansion is usually associated with increased cell density (79), and we saw the opposite (decreased cell density in high-fat diet mice). In addition, high-fat diet mice appeared to have disruptions in columnar organization, as measured by the orientation of the chondrocyte columns relative to the longitudinal axis of the bone (see Fig. 5). Although there are multiple mouse models of columnar disorganization in growth plates (80), they are primarily associated with decreased bone length and growth plate activity, highlighting another high-fat diet paradox.

It is unclear why and how the high-fat diet growth plates would be elongating faster yet have fewer, less organized proliferative cells. In some respects, the linear growth acceleration and advanced bone age in high-fat diet mice appear similar to normal and/or perhaps dysregulated growth plate aging and senescence. For instance, the decreased cell density and organization that we observed in the proliferative zone of the high-fat diet growth plates are consistent with a more senescent growth plate (81). Adipokines such as leptin can also impact growth plate structure, and there did appear to be a trend toward reduced serum leptin levels at 5 wk age (Fig. 2). Leptin-deficient obese mice have fewer growth plate columns but also reduced cell volume and overall bone length (82).

There could also be interaction of proinflammatory cytokines. VEGF was elevated in high-fat diet mice, whereas TNFα and IL-6 were reduced relative to controls. The significant increase in VEGF in the faster growing, but not yet obese, mice is not surprising given its critical role in bone elongation, remodeling, and fracture repair (83, 84). The unexpected results were the significant decreases in TNFα and IL-6 since literature evidence suggests that these markers of inflammation would be increased in the high-fat diet group (17, 85). Since elevated levels of IL-6 are known to suppress bone growth (86), our findings of decreased IL-6 are consistent with accelerated bone lengthening. Likewise, since TNFα promotes bone resorption (87), it is not surprising that we would see reduced TNFα in the same group of mice that had higher bone density.

Parallels with Pediatric Literature

The overall growth patterns in mice on a high-fat diet showed striking similarities to those described in human pediatric literature, particularly the findings that accelerated growth can precede the onset of obesity (18, 19). In a mixed-sex sample of prepubertal children, Papadimitriou et al. (88) found that those with obesity onset after 3 yr age were significantly taller than their nonobese peers when they were 2 yr old (before they had developed signs of obesity), indicating that growth acceleration preceded their obesity onset. Retrospective studies of overweight young adults have shown that advanced bone age and accelerated skeletal maturation were apparent during early childhood even before their height differences were significant (89). In fact, BMI can predict accelerated skeletal maturation as early as preschool age (12), and accelerated weight gain during the first 2 yr after birth is a significant risk factor for subsequent childhood obesity (11, 90–92). It may not be inconsequential that during a time when child and adolescent obesity rates have risen precipitously, skeletal maturation has accelerated to the point that epiphyseal fusion now occurs 7–10 mo earlier in children today compared with a century ago (93). However, not all children with obesity exhibit accelerated skeletal growth and maturation if other conditions are present (9). For example, children that do not receive adequate nutrition may suffer early growth failure followed by obesity in childhood. Although the cause of this “double burden” of growth stunting and obesity is not fully understood, it clearly has complex effects on growth physiology (94, 95).

One way our study differs from humans is that we focused on bone elongation during the most rapid period of growth that occurs in prepubertal mice, and so our model does not account for factors associated with a pubertal growth spurt in humans. However, most overweight and obese children do not exhibit a typical pubertal growth spurt. Children with obesity are taller and tend to grow faster than nonobese peers; however, they grow relatively less during their teenage years (96), attain a skeletal maturity peak when normal-weight children experience a pubertal growth spurt (89), and reach adult height sooner (10). Moreover, children with obesity have both an earlier onset and earlier completion of puberty with less pubertal height gain (97, 98), regardless of pubertal timing (99). Based on these data, we believe our mouse model is applicable to studying mechanisms of growth acceleration in humans because obesity effects on stature are most pronounced before children reach the pubertal growth period and, in many cases, before they even become obese.

Caveats and Limitations

One caveat to our study involves the diet selection, since there is debate over how well a 60% fat diet can mimic the type and amount of fat realistically consumed in a human high-fat diet (100). Speakman (100) points out that a typical American diet contains up to 40% energy from fat, so a human high-fat diet could reasonably contain up to 60% energy from fat, which would be about a 1.5-fold increase from normal fat consumption. A 60% fat content of a rodent diet, however, represents a disproportionately larger change (sixfold increase in fat) from standard rodent chow (∼10% fat) (100). Furthermore, the fat composition may not parallel the actual fatty acid ratios of a human diet (100). Bortolin et al. (101) found that Western diets (fat, sugar, and salt components that mimic the typical human diet) and cafeteria diets (human snack foods) more closely modeled human metabolic dysfunction when compared with a high-fat diet. Although we intentionally used diets that were matched in nutritional content except for fat to carbohydrate ratios, it will be important to replicate our study using different diets to confirm that bone growth acceleration can be modeled under different dietary conditions.

Another limitation of our study is that we did not have a reference population for BMI in young growing mice. Using percentile ranks from within our sample restricts the number of individuals that are classified as overweight or obese to mice with BMI greater than 85% of their age- and sex-matched cohort. This limitation has only a modest impact at the 4- and 5-wk age points when some of the larger mice were controls and mass was proportionate to length in both diets (see Supplemental Fig. S2), but the method appears to underestimate the number of overweight and obese mice at 12 wk age when body mass is substantially higher. Future studies could improve this approach by expanding the reference populations to better classify obesity in growing mice as others have similarly done for rats (45).

A related caveat is that some of the mice in the 5-wk age group were singly caged for food intake studies and we cannot rule out housing effects on the amount of weight gain. Single housing is known have a wide range of effects on stress, physiology and behavior of naturally social rodents (102, 103). Some studies have found that singly caged mice eat more and tend to become obese (104), whereas others found that single housing had unexpectedly little effects on food intake and weight gain (105). Despite housing conditions, the relevant outcome is that accelerated linear growth was reproducible in all of our experimental trials.

One surprising and initially perplexing result of our in vivo IGF-I uptake studies is that the control mice appeared to have more IGF-I in their growth plates despite lower total fluorescence levels (growth plate and vasculature combined) when compared with the faster growing high-fat diet mice (Fig. 3B). These seemingly paradoxical results might be explained, in part, by the IGF binding proteins (IGFBPs). IGFBPs are produced by growth plate chondrocytes (106–110) and can be tethered to the matrix or cell surface (111–114). IGF-I binds to IGFBPs with equal or greater affinity than to its own receptor (115). In our growth plate uptake experiments, we could not discriminate whether IGF-I was bound to its receptor or to IGFBPs. This is an important issue to address in future studies because IGF-I is biologically inactive when IGFBP-bound (115, 116), and so more IGF-I in growth plates would not necessarily translate into increased IGF-I bioactivity.

Our postmortem immunostaining results in controls (decreased pIGF-IR/IGF-I bioactivity with increased IGFBP-4/IGF-I inhibitor as shown in Fig. 4) support the idea of binding protein interference. We are working to address this issue in our ongoing experiments. We are also working on more robust measurements of serum IGF-I. Although our IGF-I data from the cytokine panel are consistent with pediatric literature (28–30), it is important to incorporate assays that are validated for measuring free and bound fractions of IGF-I (117) to obtain accurate circulating levels.

Likewise, although our results suggest that IGF-mediated growth might be involved in diet-enhanced bone lengthening, our results do not rule out other factors, such as IGF-I acting through the insulin receptor (69, 70), bone formation/resorption (118, 119), or other cytokine-mediated pathways (120–122), as well as other local systems in the growth plate such as C-type natriuretic peptide (CNP) and its receptor (123). Our primary goal here was to describe and validate a model that can be used to test these and other potential mechanisms of growth acceleration in future studies. Finally, although we did not detect significant sex by diet interactions in the skeletal variables at any age point examined here, they should not be discounted in future studies since sex-specific differences are apparent in body growth (124), bone quality and fracture toughness (125), adipose deposits (126), as well as IGF-IGFBP-driven bone acquisition and adiposity in older mice (127–131).

Summary and Conclusions

Just as the C57BL/6 mouse on a high-fat diet is considered a good model for human metabolic disorders with acknowledged limitations (75, 132–136), we believe that our data validate the utility of this model, with its limitations, for studying mechanisms of accelerated skeletal growth in children. Using a high-fat diet to mimic the development of juvenile obesity in growing mice, our results demonstrate for the first time that accelerated bone elongation in growth plates precedes excess weight gain and parallels well-documented trends in pediatric literature, which show that tall stature and advanced bone age occur in children before they actually exhibit signs of obesity. We used novel experiments that integrate in vivo imaging of IGF-I delivery to growth plates with ex vivo protein assays to reveal alterations in IGF-I uptake, activity, and disorganization of growth plates on a local level in response to a high-fat diet. These results provide necessary groundwork for future studies aimed at determining the mechanisms that drive bone growth acceleration in obesity. Of perhaps more immediate relevance, by demonstrating cellular changes in the growth plate before obesity is apparent, our study highlights the importance of identifying children with advanced bone age that may be at risk of developing obesity in order to apply appropriate dietary interventions and potentially prevent some of the ensuing musculoskeletal complications.

DATA AVAILABILITY

All relevant data generated for this study are included in the manuscript and Supplemental Material.

SUPPLEMENTAL DATA

Supplemental Methods and Figs. S1 and S2: https://doi.org/10.6084/m9.figshare.16850215.

GRANTS

This research was funded by the National Institute of General Medical Sciences (P20GM121299) and National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health (R15AR067451). The NASA West Virginia Space Grant Consortium provided a research stipend to C.A.M. and D.M. (NASA Agreement No. 80NSSC20M0055). Work was made possible by equipment purchased through the Marshall University Joan C. Edwards School of Medicine Dean’s Initiative and Marshall Health. This project was supported by resources from the WV-CTSI (NIGMS Award 2U54GM104942), WV-INBRE (NIGMS award P20GM103434), Marshall University Molecular and Biological Imaging Center, Marshall University Genomics Core, and Appalachian Center for Cellular transport in Obesity Related Disorders (ACCORD).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.L.M., C.A.W., and M.A.S. conceived and designed research; A.L.M., C.A.W., C.A.M., D.M., S.E., D.T., J.D.H., D.C., H.C., and M.A.S. performed experiments; A.L.M., C.A.W., and M.A.S. analyzed data; A.L.M., C.A.W., and M.A.S. interpreted results of experiments; M.A.S. prepared figures; M.A.S. drafted manuscript; A.L.M., C.A.W., J.D.H., H.C., and M.A.S. edited and revised manuscript; A.L.M., C.A.W., C.A.M., D.M., S.E., D.T., J.D.H., D.C., H.C., and M.A.S. approved final version of manuscript.