Ultra-processed food targets bone quality via endochondral ossification

The study investigates how ultra-processed diets affect skeletal development during the critical postnatal growth period. Long bones grow via endochondral ossification (EO) in the growth plate, where tightly regulated chondrocyte proliferation, hypertrophy, matrix production, and mineralization drive longitudinal growth. Nutrition is a key modifiable environmental factor for achieving optimal bone mass and structure before and during puberty. Modern food supplies are dominated by ultra-processed foods (UPF), with high intake among children and links to obesity and metabolic disorders, but their impact on skeletal development and bone quality had not been evaluated. The authors hypothesized that an unbalanced ultra-processed diet (UPD) during postnatal growth disrupts EO and compromises bone structure and mechanical integrity.

The introduction summarizes that EO is controlled by pathways including TGFβ/BMP, FGF, Wnt, Hedgehog, and Sox9. It notes widespread consumption of UPF in recent decades, with a substantial proportion of children's energy intake derived from such foods and added sugars, which is associated with obesity, metabolic syndrome, and diabetes. Prior studies on malnutrition or malabsorption link to growth retardation and short stature, but the cellular-level effects of excessive UPF consumption on bone development and quality had not been studied.

Design: In vivo experiments using female Sprague–Dawley rats from 3 to 9 weeks of age (weaning to puberty). Animals housed under standard conditions with ad libitum access to food and drink. Measurements included body weight, body length, food/liquid intake; femur and lumbar vertebra lengths and microarchitecture assessed by micro-computed tomography (µCT) after 3 and 6 weeks (ages 6 and 9 weeks). Mechanical testing by three-point bending on femora. Histology and in situ hybridization on tibial growth plates (GP). RNA sequencing on isolated tibial GPs. Additional dietary intervention arms tested macronutrient and micronutrient contributions and real-world eating patterns.

Primary UPF experiment: Two groups (n=8 each): Control standard rat diet vs UPF + caloric soft drink (CSD). UPF meal (hamburger roll, hamburger, tomatoes, lettuce, ketchup, French fries) homogenized into patties; CSD provided. All had ad libitum access.

Outcome assessments:

• Anthropometrics: body weight, nose-to-tail length; femur length by µCT.
• µCT: femora and lumbar vertebrae scanned (SkyScan 1174; 50 kV, 800 µA, 0.25-mm Al filter, 4500 ms exposure, 13.8 µm resolution; 900 projections, 0.4° step, 2-frame averaging, 360°). Cortical region: 200 slices, threshold 87–255; trabecular: 150 slices, adaptive threshold 85–255. Parameters: BV/TV, Tb.N, Tb.Th, Tb.Sp; cortical Ct.Ar/Tt.Ar, Ct.Th, Ma.Ar, BMD; porosity Ct.Po, Po.N, Po.V.
• Mechanics: Three-point bending; derived stiffness, yield load, maximum load, fracture load, energy to fracture from load–displacement curves.
• Histology: H&E, Safranin O, Masson’s trichrome, TRAP. In situ hybridization for Col2 and Col10. Growth plate zone widths (PZ, HZ, lesion zone LZ) measured at 10 locations per GP; percentages calculated.
• SEM: high-resolution imaging of GP surface.
• Serum analyses: hematological and hormonal profiles including Ca, phosphate, PTH, FGF23, OPG; metabolic status.

Transcriptomics: Tibial GP isolated at 3 weeks on diet; RNA extracted from pulverized GP. High-quality RNA (RIN>7) selected. Libraries prepared with QuantSeq 3' mRNA-Seq (Lexogen); sequenced on Illumina NextSeq 500 (single-end, ~25M reads/sample). Libraries: Control n=4, UPF n=5 (from initial n=8 per group). QC with FastQC; trimming with Trim Galore; mapping to rat genome (Rnor_6.0.94) with STAR; processing with Samtools; counts with HTSeq-count; DE analysis with DESeq2 (adjusted P<0.05; |fold change|>1). Gene set analyses with GeneAnalytics after orthology mapping (OMA, Biomart). VEGF, ECM, MMP, ADAM families examined.

Macronutrient experiment: Four groups: Control (n=16), UPF+CSD (n=18), high-fat via corn oil addition (Corn; n=16), high-sucrose via 10% sucrose solution plus control diet (Sucrose; n=16). Assessed µCT, GP histology and zone ratios.

Micronutrient experiment (Ca/P): Three groups: Control (n=20), UPF without CSD (n=20), custom diet with Ca 0.06% (0.62 mg/g) and P 0.12% (1.21 mg/g) mimicking UPF Ca:P ratio; other nutrients matched control (n=8). Assessed growth, intake of Ca/P, serum markers, µCT, mechanics, GP histology.

Eating pattern experiment: Three groups (n=8 each): Control, UPF, and 30%/70% group (30% of week on Control diet, remainder on UPD) to mimic children’s UPF consumption patterns. Assessed intake, growth, µCT, porosity, mechanics, GP histology.

Statistics: Mean ± SD; one-way ANOVA with Tukey–Kramer HSD and t-tests; significance at P≤0.05.

• Growth: UPF+CSD rats showed reduced weight gain and significantly shorter total body and femoral lengths despite higher total caloric intake, indicating growth retardation not due to caloric deficiency.
• Trabecular bone: Femoral BV/TV decreased markedly in UPF+CSD vs Control (e.g., from 35.54% to 19.17% at 6 weeks; from 23.06% to 16.27% at 9 weeks). Tb.N and Tb.Th were significantly lower, and Tb.Sp was higher at both time points. Vertebral trabecular parameters showed similar impairment.
• Cortical bone: UPF+CSD exhibited decreased Ct.Ar/Tt.Ar and approximately 30% lower BMD vs Control. Cortical porosity was greatly increased: Po.N 15-fold higher and Po.V 47-fold higher. Cross-sections showed a sieve-like cortex.
• Mechanics: Three-point bending demonstrated large reductions in stiffness, yield, maximum, and fracture loads in UPF+CSD bones; load–displacement curves showed pronounced weakening.
• Growth plate (GP): Histology revealed uneven GP widening with avascular, nonmineralized cartilage lesions extending into metaphysis (lesion zone, LZ). Proliferative zone organization was disrupted; LZ cells lost columnar alignment and hypertrophic characteristics. In situ hybridization showed strong Col2 in PZ and Col10 above plaques but absence of Col10 in lesion extensions, indicating loss of hypertrophic properties. Mineralization at the cartilaginous plaque was absent.
• Transcriptomics (GP): Of 12,848 genes, 302 were differentially expressed (74% upregulated in UPF). ECM synthesis genes were upregulated (e.g., COMP, Acan, Col2, Col4, Col9, Col10), with no corresponding increase in MMP or ADAM family expression, indicating imbalance favoring matrix accumulation. BMP pathway inhibitors (Fst, Nog, Gdf10, Inhba) were upregulated, suggesting attenuated BMP signaling. Sox9 was upregulated along with downstream targets (Acan, Col2, Col10), indicating altered proliferation/differentiation. Mineralization-related genes Dmp1 and Phex were downregulated and MGP upregulated, consistent with impaired mineralization. Vascularization genes (vegfa, vegfb, vegfc, PECAM-1) were not differentially expressed.
• Serum/metabolic and osteoclast activity: UPF group showed reduced FGF23, hyperphosphatemia, hypocalcemia, elevated PTH, and reduced OPG, consistent with increased bone resorption; cortical TRAP staining showed elevated osteoclast activity and increased cortical porosity, while osteoclast activity at the chondro-osseous junction was not different.
• Macronutrient arms: High-fat (Corn) or high-sucrose diets alone did not reproduce the trabecular/cortical deficits or GP lesions; bone parameters and GP organization remained normal.
• Micronutrient Ca/P arm: The Ca/P group (low Ca, high P ratio mimicking UPF) exhibited growth retardation and µCT patterns similar to or worse than UPF, with significantly worsened bone mechanical properties (e.g., femoral stiffness ~3× lower than UPF). GP lesions occurred but were milder and less frequent (42% of GPs) than UPF (86% of GPs); GP expansion was reduced vs UPF. Some serum mineral parameters were better balanced than UPF, suggesting compensation at the expense of bone.
• Eating pattern arm (30% Control/70% UPD): Despite lower caloric and mineral intake than UPF, growth patterns resembled Control. However, trabecular parameters (except Tb.Th) were similar to UPF; cortical parameters improved relative to UPF but did not reach Control; mechanical properties were improved vs UPF but remained inferior to Control. GP organization was similar to Control. Even partial UPD exposure impaired skeletal quality.

UPD consumption during growth severely disrupts skeletal development by targeting the EO process in the growth plate, leading to stunted longitudinal growth and compromised bone quality and strength. The hallmark lesion is an avascular, noncalcified cartilage plaque extending from the GP into the metaphysis with disorganized chondrocyte arrangement and disrupted zone proportions. GP transcriptomics indicate a mechanistic basis: increased ECM production without matching degradation (unchanged MMP/ADAM), attenuated BMP signaling via upregulated inhibitors, and elevated Sox9 with downstream matrix gene expression, all consistent with impaired transitions in chondrocyte proliferation and hypertrophy. Mineralization defects are supported by downregulation of Dmp1 and Phex and upregulation of MGP, aligning with decreased mineral deposition. While Dmp1/Phex knockout models classically elevate FGF23 causing hypophosphatemia, here FGF23 decreased with hyperphosphatemia and hypocalcemia, alongside elevated PTH and reduced OPG, pointing to enhanced cortical osteoclast-mediated resorption and increased cortical porosity. The phenotype is not explained by caloric intake or by isolated macronutrients (fat or sugar) or the soft drink alone. Ca/P deficiency and imbalance reproduces part of the phenotype (growth retardation, poor bone quality, GP changes) but is less severe and likely acts via partly different mechanisms. This suggests additional UPD-related cofactors, potentially linked to food processing, synergize with micronutrient imbalance to impair EO and bone quality. The findings identify bone as a critical, previously underappreciated target of ultra-processed diets during growth, with implications for fracture risk and peak bone mass attainment.

Feeding growing rats an ultra-processed diet profoundly impairs bone development: it disrupts growth plate organization and EO, reduces trabecular and cortical bone quality, increases cortical porosity, lowers mineral density, and weakens mechanical strength, thereby elevating fracture risk. GP transcriptomics reveal ECM accumulation, attenuated BMP signaling, Sox9 upregulation, and impaired mineralization as key mechanisms. The deleterious skeletal effects are not due to caloric deficit, isolated high fat or sugar, or the soft drink alone. Calcium/phosphorus deficiency and imbalance accounts for part of the phenotype but additional UPD-related factors likely contribute. Even partial UPD exposure (30%/70% pattern) degrades bone quality despite normal growth metrics. These results highlight bone as a new target of modern ultra-processed diets and motivate future studies to identify specific processing-related factors and interventions to protect the developing skeleton.

The study shows that Ca/P imbalance partially mimics the UPF phenotype but explicitly notes that the same mechanisms cannot be concluded to underlie both conditions and that other cofactors in UPD are likely involved but were not identified. Soft drink and isolated macronutrients were ruled out as sole causes, but specific additional detrimental components of UPD were not delineated.