Ganagliflozin proline attenuates osteogenic dysfunction and inflammatory signaling during exposure to advanced glycation end products in murine osteoblasts

Abstract

BACKGROUND

Advanced glycation end products (AGEs) accumulate in diabetes, disrupt the collagen I-rich bone matrix, activate inflammatory signaling involving the receptor for AGEs (RAGE) and nuclear factor kappa B (NF-κB), and impair osteoblast function. Ganagliflozin proline (Gana Pro), a sodium-glucose cotransporter 2 inhibitor, may protect bone, but its effects on AGE-induced osteogenic dysfunction remain unclear.

AIM

To determine whether Gana Pro attenuates AGE-induced osteogenic dysfunction and examine associated collagen I and inflammatory signaling changes.

METHODS

AGE accumulation was measured by enzyme-linked immunosorbent assay. Molecular docking assessed potential AGE-collagen I interactions, and collagen I-methylglyoxal co-incubation assessed modification. AGE-treated MC3T3-E1 cells received canagliflozin or Gana Pro. Cell viability, proliferation, cell-cycle distribution, collagen I, inflammatory signaling, osteogenic markers, alkaline phosphatase activity, and mineralization were evaluated using molecular, cellular, imaging, and staining assays. Separate RAGE-knockdown experiments with phorbol 12-myristate 13-acetate (PMA) stimulation assessed pathway involvement.

RESULTS

Higher AGE levels were observed under 25 mmol/L glucose than under 5.5 mmol/L glucose. Exploratory docking predicted non-covalent poses of representative AGE-related molecules within a fibronectin-collagen peptide complex, while methylglyoxal increased AGE-related fluorescence in a cell-free collagen I system. Commercial AGE-bovine serum albumin (BSA) exposure was associated with reduced MC3T3-E1 viability, proliferation, collagen I, osteogenic markers, alkaline phosphatase activity, and mineralization, together with increased RAGE, p65 phosphorylation, and tumor necrosis factor alpha (TNF-α). Gana Pro, particularly at 5 μM, improved proliferation, collagen I and osteogenic marker expression, alkaline phosphatase activity, calcium deposition, and mineralization, while reducing RAGE, p65 phosphorylation, and TNF-α. Separately, RAGE knockdown reduced AGE-BSA-associated p65 phosphorylation and restored osteogenic markers, whereas PMA partially attenuated this recovery.

CONCLUSION

Gana Pro attenuated AGE-induced osteogenic dysfunction, accompanied by restored collagen I and reduced RAGE/NF-κB inflammatory signaling; direct pharmacological rescue is required to establish pathway mediation.

Key Words: Advanced glycation end products; Collagen I; Diabetic bone fragility; Ganagliflozin proline; Nuclear factor kappa B; Osteogenic differentiation; Receptor for advanced glycation end products; Sodium-glucose cotransporter 2 inhibitor

Core Tip: This study demonstrates that ganagliflozin proline (Gana Pro) protects MC3T3-E1 osteoblasts against advanced glycation end product-induced dysfunction. Gana Pro improved proliferation, collagen I expression, osteogenic-marker expression, alkaline phosphatase activity, calcium deposition, and matrix mineralization. These benefits were accompanied by lower receptor for advanced glycation end products (RAGE) expression, p65 phosphorylation, and tumor necrosis factor alpha expression. A separate RAGE-knockdown/phorbol 12-myristate 13-acetate experiment supported the functional involvement of this signaling axis in advanced glycation end products-induced osteogenic suppression. The combined evidence identifies RAGE/nuclear factor kappa B signaling as a plausible mechanistic contributor, although drug-specific pathway mediation requires direct rescue validation.

INTRODUCTION

Type 2 diabetes mellitus (T2DM) is increasingly recognized as a systemic disorder whose skeletal consequences extend beyond loss of bone mass alone[1]. Patients with T2DM have greater hip-fracture risk despite normal or elevated bone mineral density (BMD), a phenomenon termed the diabetic bone paradox[2,3]. A meta-analysis confirmed higher hip-fracture risk alongside increased spinal and hip BMD Z-scores, implicating impaired bone quality and remodeling rather than BMD alone[4]. Advanced glycation end products (AGEs), formed through non-enzymatic reactions between reducing sugars and proteins or lipids, link chronic hyperglycemia to skeletal deterioration[5]. Their deposition increases in diabetic bone matrix, where abnormal collagen crosslinking promotes stiffness, microdamage, and mechanical failure[6,7]. AGE-related biomarkers correlate more consistently with fractures than with BMD, suggesting that densitometry may underestimate glycation-associated skeletal fragility[8,9].

Collagen I (COL1), the principal organic component of bone, provides the scaffold for mineral deposition and contributes substantially to tissue toughness[10]. Its long half-life makes it particularly susceptible to cumulative non-enzymatic glycation[11]. AGE modification disrupts fibril organization, restricts intermolecular sliding, stiffens the matrix, and reduces energy dissipation[12,13]. Computational models further link increasing AGE crosslink density to abrupt fibril fracture, providing a nanoscale explanation for diabetic bone weakness[12]. Complementing these material findings, Park et al[14] reported that AGEs reduced alkaline phosphatase (ALP) activity and osteoblast-related gene expression and tended to reduce bone-nodule formation in MC3T3-E1 cells.

AGEs also function as bioactive ligands that elicit intracellular stress responses[15,16]. The receptor for AGEs (RAGE) is a pattern-recognition receptor central to diabetes-associated inflammation[17]. AGE–RAGE engagement can activate mitogen-activated protein kinase and nuclear factor kappa B (NF-κB) signaling, promoting oxidative stress, inflammatory cytokine production, apoptosis, and osteoblast dysfunction[18]. Experimental inhibition of AGE accumulation or AGE/RAGE/NF-κB signaling can improve osteoblast viability, osteogenic differentiation, and bone-remodeling homeostasis in diabetic osteoporosis models[19,20]. Separately, Li et al[21] found that RAGE overexpression inhibited osteoblast proliferation, whereas RAGE neutralization restored it. That study implicated Wnt, phosphoinositide 3-kinase, and extracellular signal-regulated kinase signaling, not NF-κB. Matrix glycation and RAGE-associated cellular responses thus offer complementary, but not necessarily identical, explanations for AGE-induced osteogenic impairment[22]. Whether one pharmacological intervention can improve both dimensions while restoring osteoblast function remains unresolved.

Sodium-glucose cotransporter 2 (SGLT2) inhibitors have reshaped T2DM treatment through glycemic, cardiovascular, and renal benefits, but their skeletal effects remain uncertain[23-26]. Early studies raised concerns about total-hip BMD loss during canagliflozin (Cana) treatment[27]. Later analyses did not consistently identify a class-wide fracture signal, and meta-analytic evidence suggests no overall increase in fractures or reduction in BMD[28]. These divergent findings suggest context- and compound-specific skeletal actions. Experimental osteogenic activity reported for Cana provides a pharmacological precedent, not evidence of a uniform class effect[29]. Ganagliflozin proline (Gana Pro), a proline cocrystal of janagliflozin, is an SGLT2 inhibitor whose active parent has undergone clinical study[30]. However, neither this clinical background nor related-compound evidence demonstrates that Gana Pro protects osteoblasts or targets RAGE/NF-κB signaling. Its effects on AGE-induced osteogenic injury therefore require direct assessment.

Accordingly, the primary objective of this study was to determine whether Gana Pro attenuates AGE-induced deficits in osteoblast proliferation, differentiation, and mineralization. We also examined AGE-related COL1 modification to connect extracellular matrix injury with cellular dysfunction. The secondary objective was to assess whether Gana Pro-associated protection was accompanied by changes in COL1 and RAGE/NF-κB/tumor necrosis factor alpha (TNF-α) signaling. To examine the functional contribution of this pathway to AGE injury, we conducted a separate pathway-perturbation experiment using short interfering RNA targeting RAGE and phorbol 12-myristate 13-acetate (PMA). PMA was used as an NF-κB-activating stimulus, but Gana Pro was not included in this experiment. This design can support pathway involvement in AGE-induced dysfunction but cannot establish that RAGE/NF-κB signaling mediates the specific effects of Gana Pro.

MATERIALS AND METHODS

Reagents and antibodies

A commercial AGE-modified bovine serum albumin preparation (AGE-BSA) (10 mg; bs-1158P; Bioss, Beijing, China) was used for cellular exposure experiments. According to the manufacturer, AGE-BSA was produced by reacting bovine serum albumin with glycolaldehyde under sterile conditions, followed by extensive dialysis and purification. The manufacturer specifies 0.01 mol/L phosphate-buffered saline (PBS) as the formulation buffer and greater than 95% protein purity by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. This specification describes protein purity and does not quantify AGE content or the degree of glycation. All treatment concentrations refer to the total mass concentration of the AGE-BSA preparation. Batch-specific AGE content, glycation degree, and residual glycolaldehyde concentration were not reported. The manufacturer lists the endotoxin status as not analyzed. These characteristics were not independently quantified, and a matched unmodified bovine serum albumin control was not included in the original experimental design.

Gana Pro (batch 010103-2504016) was characterized in a certificate of analysis issued by Jilin Huikang Pharmaceutical Co., Ltd. Identity relative to the reference material was confirmed by chromatographic retention time, infrared spectroscopy, and X-ray powder diffraction. The certificate reported 66.0% janagliflozin active moiety and 33.2% L-proline on an anhydrous basis, together with 1.5% water and 0.38% total related substances. These component values are reported separately and are not equivalent to a statement of 99.2% chemical purity.

Cana (M2002; AbMole, Shanghai, China; manufacturer-reported purity, 99.80%), methylglyoxal (MGO) (M109035; Aladdin, China), aminoguanidine hydrochloride (A151036-25 g; Aladdin, China), purified type I collagen from rat tail (40125ES10; Yeasen, China), and PMA (HY-18739; MedChemExpress, United States) were obtained commercially. Stock solutions of Gana Pro and Cana were prepared in dimethyl sulfoxide and diluted in culture medium immediately before use. The final dimethyl sulfoxide concentration was matched among solvent-containing groups and did not exceed 0.1% (v/v).

Primary antibodies and horseradish peroxidase-conjugated secondary antibodies were obtained from Proteintech or Bioss, as detailed in Supplementary Table 1. Antibody catalogue numbers, host species, working dilutions, and manufacturer-stated molecular weights are provided in Supplementary Table 2.

Cell culture

Mouse MC3T3-E1 preosteoblasts were obtained from Haixing Biosciences (TCM-C817, China). According to the supplier’s quality-control documentation, the cells were authenticated and tested negative for mycoplasma contamination before distribution. Cells were cultured in α-minimum essential medium (G4555; Servicebio, China) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin and maintained at 37 °C in a humidified atmosphere containing 5% carbon dioxide. The medium was replaced every 2-3 days, and cells at passages 5-15 were used. Normal-glucose (NG) and high-glucose (HG) conditions contained 5.5 mmol/L and 25 mmol/L glucose, respectively. Cellular injury experiments used commercial AGE-BSA, principally at 200 μg/mL total AGE-BSA protein, and the main Gana Pro concentration was 5 μmol/L. All cellular findings were obtained using MC3T3-E1 cells.

Experimental design and grouping

This study comprised four complementary experimental modules. First, AGE levels were compared between MC3T3-E1 cells cultured under NG and HG conditions for 3 days, 7 days, or 14 days. AGE levels were measured in culture supernatants and cell lysates. Because no osmotic control was included, this module was designed to characterize differences between the two culture conditions rather than establish a glucose-specific effect. An independent cell-free COL1-MGO system was used to assess AGE-related fluorescence arising during COL1 exposure to MGO. Gana Pro was not included in this system.

Second, concentration-screening experiments evaluated AGE-BSA and Gana Pro conditions. Gana Pro alone and Cana alone were tested at 1 μmol/L, 5 μmol/L, 10 μmol/L, 20 μmol/L, and 40 μmol/L in the cell counting kit 8 (CCK-8) assay, whereas AGE-BSA plus Gana Pro combinations contained 1 μmol/L, 5 μmol/L, or 10 μmol/L Gana Pro. These experiments assessed concentration-associated changes in cellular metabolic activity. A Gana Pro-only group was not included in the subsequent molecular, inflammatory, differentiation, or mineralization experiments.

Third, treatment-associated effects were evaluated in the NG, HG, AGE-BSA, AGE-BSA plus Cana, and AGE-BSA plus Gana Pro groups. The Cana concentration was 10 μmol/L. Molecular endpoints included AGE-modified proteins, COL1, RAGE, total p65, phosphorylated p65, TNF-α, ALP, runt-related transcription factor 2 (RUNX2), and osteocalcin (OCN). The phosphorylated-p65/total-p65 ratio was used as a whole-cell readout of NF-κB-associated signaling. Functional endpoints included 5-ethynyl-2’-deoxyuridine (EdU) incorporation, RUNX2 immunofluorescence, ALP staining and activity, Alizarin Red S (ARS) staining, and calcium quantification.

Fourth, the functional contribution of RAGE/NF-κB-associated signaling to AGE-BSA-associated injury was examined in the NG plus negative-control small interfering RNA (siRNA), AGE-BSA plus negative-control siRNA (siNC), AGE-BSA plus RAGE-targeting siRNA (siRAGE), and AGE-BSA plus RAGE-targeting siRNA plus PMA groups. This pathway-interrogation module was conducted independently of the Gana Pro experiments and did not contain a Gana Pro plus PMA group. It therefore assessed pathway involvement during AGE-BSA exposure but did not test whether this pathway mediated the effects of Gana Pro.

Enzyme-linked immunosorbent assay measurement of AGE accumulation

MC3T3-E1 cells were cultured under NG or HG conditions for 3 days, 7 days, or 14 days. At each time point, culture supernatants were collected and centrifuged at 1000 × g for 10 minutes at 4 °C. Cells were washed with chilled PBS, lysed in radioimmunoprecipitation assay buffer containing protease inhibitors, and centrifuged at 12000 × g for 15 minutes at 4 °C.

AGE levels were measured using a competitive general advanced glycation end product enzyme-linked immunosorbent assay kit (BSKL64554; Bioss, China). The assay range was 125-8000 ng/mL, and the manufacturer-reported sensitivity was 44 ng/mL. Absorbance was measured at 450 nm, and concentrations were calculated from the standard curve. Each sample was measured in 3 technical wells, which were averaged before statistical analysis. The number and definition of independent experimental replicates are reported in the corresponding figure legend. These comparisons lacked an osmotic control and were therefore interpreted as differences between the specified culture conditions.

Exploratory non-covalent docking within a fibronectin-collagen peptide complex

Exploratory docking assessed possible non-covalent poses of pentosidine (PubChem CID 119593), Nε-(carboxymethyl)lysine (CID 123800), glucosepane (CID 135565852), and methylglyoxal-derived hydroimidazolone-1 (MG-H1) (SID 355416216; associated CID 132601822). The receptor was derived from the crystal structure of a low-affinity collagen site in complex with fibronectin type I domains 8 and 9 (Protein Data Bank accession code: 3GXE), a 2.60-Å X-ray structure containing human fibronectin domains bound to short collagen α1(I)-derived peptides.

Receptor and ligand structures were prepared using AutoDockTools 1.5.7. Water molecules and nonessential heteroatoms were removed, polar hydrogens were added, partial charges and atom types were assigned, and ligand torsions were defined. The prepared receptor and ligand structures were saved in PDBQT format for docking. Docking was conducted using AutoDock Vina 1.1.2 with a rigid receptor and flexible ligands. The search box was centered at x = 40.974, y = -14.838, and z = -20.361 Å, with dimensions of 54.706 Å × 62.339 Å × 62.339 Å. Exhaustiveness was 8, the energy range was 3 kcal/mol, and nine poses were generated per ligand.

Each ligand underwent one docking run, and the generated poses were not considered independent replicates. Top-ranked poses were visualized using PyMOL 2.6.0 and Discovery Studio Client 4.5. No redocking, replicate-run analysis, benchmarking, covalent docking, molecular-dynamics simulation, or experimental binding validation was performed.

Cell-free COL1-MGO co-incubation and AGE-related fluorescence detection

Purified rat-tail COL1 (40125ES10; Yeasen, China) was diluted to 1 mg/mL in sterile 1 × PBS (potential of hydrogen = 7.4). A 50-μL aliquot was added to each well and allowed to gel at 37 °C for 30 minutes. Each collagen gel was then overlaid with 100 μL of PBS containing 10 mmol/L MGO (M109035; Aladdin, China), with or without 20 mmol/L aminoguanidine hydrochloride. The final assay volume was 150 μL/well, and samples were incubated at 37 °C for 5 days. The groups were COL1, MGO, COL1 + MGO, and COL1 + MGO + aminoguanidine hydrochloride. In the MGO-only group, 50 μL of PBS replaced the collagen gel. MGO alone served as a fluorescence-background control, whereas aminoguanidine served as an inhibitor of AGE formation. Fluorescence was measured at excitation and emission wavelengths of 370 nm and 440 nm, respectively. Values were normalized to the mean fluorescence of the COL1 group, which was set to 1.0.

Cell viability assay

Cellular metabolic activity was assessed using a CCK-8 assay (C0038; Beyotime, China). MC3T3-E1 cells were seeded in 96-well plates at 3000 cells per well and allowed to adhere overnight. Injury-condition screening used AGE-BSA at 25 μg/mL, 50 μg/mL, 100 μg/mL, 200 μg/mL, or 400 μg/mL for 12 hours, 24 hours, or 48 hours. Gana Pro alone and Cana alone were evaluated at 1 μmol/L, 5 μmol/L, 10 μmol/L, 20 μmol/L, or 40 μmol/L over the same time range. Intervention screening used 200 μg/mL AGE-BSA combined with 1 μmol/L, 5 μmol/L, or 10 μmol/L Gana Pro. At each endpoint, CCK-8 reagent was added, and cells were incubated at 37 °C for 1.5 hours. Absorbance was measured at 450 nm. Technical wells were averaged to obtain one value for each independent experiment. On the basis of the combined concentration-screening results, 200 μg/mL AGE-BSA and 5 μmol/L Gana Pro were selected as the working concentrations.

EdU incorporation assay

Cell proliferation was evaluated using an EdU assay kit (GDP1023; Servicebio, China). MC3T3-E1 cells were exposed to 200 μg/mL AGE-BSA alone or together with 1 μmol/L, 5 μmol/L, or 10 μmol/L Gana Pro for 24 hours. Cells were subsequently incubated with the EdU working solution according to the manufacturer’s instructions. The cells were fixed with 4% paraformaldehyde, permeabilized with 0.3% Triton X-100, and subjected to click-reaction staining. Nuclei were counterstained with Hoechst 33342 (S0004; Bioss, China).

Fluorescence images were acquired under identical exposure and magnification settings. An equal number of non-overlapping fields was analyzed from each technical replicate well using ImageJ. The same image-analysis parameters and threshold range of 35-255 were applied across all experimental groups. The percentage of EdU-positive cells was calculated by dividing the number of EdU-positive nuclei by the total number of Hoechst-positive nuclei in the corresponding field and multiplying by 100. The EdU-positive area ratio was calculated by dividing the EdU-positive percentage area by the corresponding Hoechst-positive percentage area.

Three independent biological experiments were performed using separately cultured and treated cell preparations. Each biological experiment included three technical replicate wells per treatment condition. Field-level measurements were first averaged within each technical well. The three well-level values were then averaged to generate one value for each independent biological experiment. Statistical analysis was performed using the three biological-replicate values (n = 3); technical wells and imaging fields were not treated as independent observations.

Flow cytometric analysis of cell-cycle distribution

Cell-cycle distribution was assessed using propidium iodide staining solution (C1008S; Beyotime, China). Following treatment, cells were collected, washed twice with chilled PBS, fixed overnight in 70% ethanol at 4 °C, and incubated with RNase A and propidium iodide for 30 minutes at room temperature in the dark.

Data were acquired using a BD FACSCalibur flow cytometer with CellQuest Pro 6.0 software. Between 10717 and 13542 events were recorded per sample. Raw flow-cytometry standard files were reanalyzed in FlowJo. Cellular events were identified using forward- and side-scatter parameters to exclude debris, followed by singlet discrimination using propidium iodide area vs width. Cell-cycle modeling was applied to the propidium iodide-area histogram of the singlet population to estimate the proportions of cells in the G0/G1, S, and G2/M phases. The same gating hierarchy was applied to all samples.

Western blotting

Whole-cell protein was extracted using radioimmunoprecipitation assay buffer containing protease and phosphatase inhibitors. Protein concentration was determined using a bicinchoninic acid assay. Equal amounts of protein were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes. Membranes were blocked with 5% non-fat milk for 1 hour and incubated overnight at 4 °C with the indicated primary antibodies.

Membranes were subsequently incubated with horseradish peroxidase-conjugated secondary antibodies for 1 hour at room temperature. Bands were visualized using enhanced chemiluminescence reagent and quantified using ImageJ. Target proteins were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Phosphorylated p65 was evaluated relative to GAPDH and as the phosphorylated-p65/total-p65 ratio. These whole-cell measurements were interpreted as NF-κB-associated readouts; nuclear p65 translocation and nuclear/cytoplasmic fractionation were not performed. Quantification was based on n = 3 per group. Complete uncropped parent blot images, molecular-weight markers, lane identities, and panel assignments for Figures are provided in Supplementary Figures 1-3.

RNA extraction and quantitative real-time polymerase chain reaction

Total RNA was extracted using an RNA extraction kit (G3689, Servicebio, China). RNA concentration and purity were assessed using absorbance at 260 nm and the A260/A280 ratio, respectively. Complementary DNA was synthesized using a reverse transcription kit (G3331-100, Servicebio, China) according to the manufacturer’s instructions. Quantitative real-time polymerase chain reaction (RT-qPCR) was performed using SYBR Green quantitative PCR Master Mix on a real-time PCR system. The target genes included Col1a1, Ager, Rela, Tnf, Alpl, Runx2, and Bglap. Gapdh was used as the internal reference gene. Relative messenger RNA (mRNA) expression was calculated using the 2^-ΔΔCt method. Primer sequences are listed in Supplementary Table 3.

RUNX2 immunofluorescence staining

RUNX2 immunofluorescence staining was performed to evaluate osteogenic transcriptional activity. After treatment, MC3T3-E1 cells grown on coverslips were fixed with 4% paraformaldehyde for 15 minutes, permeabilized with 0.3% Triton X-100 for 10 minutes, and blocked with 5% bovine serum albumin at room temperature for 1 hour. Cells were incubated with a primary antibody against RUNX2 overnight at 4 °C, followed by incubation with a fluorescence-conjugated secondary antibody for 1 hour at room temperature in the dark. Nuclei were counterstained with 4’,6-diamidino-2-phenylindole (C0065, Solarbio, China). Fluorescence images were acquired using a fluorescence microscope or confocal microscope under identical imaging settings. RUNX2 fluorescence was quantified as the mean gray value within the analyzed regions using ImageJ. Relative fluorescence intensity was calculated by dividing each mean gray value by that of the corresponding NG sample.

Osteogenic induction

For osteogenic differentiation experiments, MC3T3-E1 cells were cultured in osteogenic induction medium. The osteogenic induction medium consisted of complete alpha minimum essential medium supplemented with 10 mmol/L β-glycerophosphate, 50 μg/mL ascorbic acid, and 100 nM dexamethasone. During induction, cells were treated with 200 μg/mL AGE-BSA, 10 μmol/L Cana, or 5 μmol/L Gana Pro according to the experimental grouping. The medium was replaced every 2-3 days, and the corresponding treatments were replenished simultaneously.

Samples for the analysis of osteogenesis-related protein and mRNA expression were collected on day 14 of osteogenic induction. Early osteogenic differentiation was evaluated by ALP staining and ALP activity assays on day 14, whereas late-stage mineralization was assessed by ARS staining, ARS-positive area analysis, and calcium quantification on day 21.

ALP staining and ALP activity assay

After osteogenic induction for the indicated duration, cells were washed with PBS and fixed with 4% paraformaldehyde. ALP staining was performed using an ALP staining kit according to the manufacturer’s instructions. Low- and high-magnification images were captured under a light microscope. The ALP-positive staining area was quantified using ImageJ. For the ALP activity assay, cells were lysed and ALP activity was measured using a colorimetric ALP activity assay kit. Absorbance was measured at the wavelength specified in the kit instructions. ALP activity in cell lysates was calculated according to the manufacturer’s instructions and expressed as U/L. This assay was used to evaluate early osteogenic differentiation.

ARS staining and calcium quantification

Matrix mineralization was evaluated using ARS staining. After osteogenic induction for the indicated duration, cells were washed with PBS, fixed with 4% paraformaldehyde, and stained with ARS solution (G1038, Servicebio, China). Excess dye was removed by washing with distilled water. Mineralized nodules were observed and imaged using a light microscope.

Calcium deposition was quantified using a calcium assay kit (S1063S, Beyotime, China) according to the manufacturer’s instructions. Calcium content was reported as μg/well and was not normalized to total protein, DNA content, or viable cell number. The ARS-positive area was expressed as a percentage of the analyzed field area. These measurements were used as functional readouts of late-stage osteogenic mineralization. In the present dataset, ALP staining/activity assays and ARS staining/calcium quantification were used as functional endpoints to support Gana Pro-mediated restoration of osteogenic differentiation and mineralization.

siRNA-mediated RAGE knockdown and PMA stimulation

To assess the functional contribution of RAGE/NF-κB-associated signaling to AGE-induced osteoblast injury, MC3T3-E1 cells were transfected with siRAGE or siNC (GenePharma Co., Ltd., Shanghai, China). The siRNA sequences and working concentration of 50 nM are provided in Supplementary Table 4. Transfection was performed using Lipofectamine RNAiMAX transfection reagent (13778150; Invitrogen, Thermo Fisher Scientific, United States) according to the manufacturer’s instructions. Briefly, the siRNA and transfection reagent were separately diluted in Opti-MEM I Reduced Serum Medium (31985070; Gibco, Thermo Fisher Scientific, United States), combined, and incubated at room temperature to allow transfection-complex formation before being added to the cultures.

At 48 hours after transfection, cells were exposed to AGEs at 200 μg/mL for 24 hours. Where indicated, PMA (HY-18739; MedChemExpress, United States) was added simultaneously with AGEs at a final concentration of 100 nM and maintained for the entire 24-hour exposure period. The experimental groups were NG + siNC, AGEs + siNC, AGEs + siRAGE, and AGEs + siRAGE + PMA.

RAGE knockdown was verified by RT-qPCR measurement of Ager mRNA and Western blot analysis of RAGE protein. Total p65, phosphorylated p65, TNF-α, COL1, OCN, ALP, and RUNX2 were subsequently evaluated at the mRNA and/or protein levels using the methods described above. The phosphorylated-p65/total-p65 ratio was used as the principal readout of NF-κB-associated signaling.

PMA was used here as an NF-κB-activating stimulus rather than as a selective NF-κB agonist. This experiment was designed to assess the functional contribution and putative signaling order of the RAGE/NF-κB axis in AGE-induced injury. It did not directly test whether NF-κB reactivation reverses the effects of Gana Pro.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 11 (GraphPad Software, Boston, MA, United States). Exact sample sizes and experimental units are defined in the corresponding figure legends. Unless otherwise specified, n = 3 denotes three independent biological experiments using separately cultured and treated cell preparations on different occasions. Each experiment included three technical replicate wells where applicable. Multiple wells, microscopic fields, or repeated instrument readings from the same experiment were treated as technical subsamples and averaged to produce one value per biological replicate. Technical subsamples were not counted as independent observations. Data are presented as the mean ± SD. Two-group comparisons in the AGE enzyme-linked immunosorbent assay datasets were performed using two-tailed Welch’s t tests. Comparisons among more than two groups were performed using ordinary one-way analysis of variance followed by Tukey’s multiple-comparisons test. For the concentration-screening experiments, treatment groups were compared separately within each experimental series at each time point using ordinary one-way analysis of variance followed by Tukey’s multiple-comparisons test. AGE enzyme-linked immunosorbent assay experiments containing glucose condition and time as separate factors were analyzed using ordinary two-way analysis of variance followed by Tukey’s multiple-comparisons test. The two-way model for the AGE enzyme-linked immunosorbent assay data included the interaction between glucose condition and time. Homogeneity of variance for one-way analyses was assessed using the Brown-Forsythe test. A two-sided P value below 0.05 was considered statistically significant. Statistical significance in the figures was denoted using superscript lowercase letters: ^aP < 0.05, ^bP < 0.01, ^cP < 0.001, and ^dP < 0.0001.

RESULTS

AGE levels under normal- and high-glucose culture conditions and AGE-related type I collagen modification

A high-glucose microenvironment has been associated with AGE formation through non-enzymatic glycation, glycoxidation, and dicarbonyl stress, which may contribute to AGE-related COL1 modification and abnormalities of the collagen matrix (Figure 1A). To compare AGE levels under different glucose conditions, cells were cultured under NG (5.5 mmol/L) or HG (25 mmol/L) conditions for 3 days, 7 days, and 14 days, and total AGE levels in the culture supernatants and cell lysates were measured. AGE levels were significantly higher under the 25 mmol/L glucose condition than under the 5.5 mmol/L glucose condition in both culture supernatants and cell lysates at all examined time points (Figure 1B and C). Time-course analysis further showed that AGE levels gradually increased with culture duration, with the highest measured AGE levels observed in the cell lysate fraction under the 25 mmol/L glucose condition on day 14 (Figure 1D-G). Under the 5.5 mmol/L glucose condition, AGE levels in the culture supernatants were higher than those in cell lysates at early time points, whereas under the 25 mmol/L glucose condition, relatively higher AGE levels were observed in the cell-associated fraction. Overall, the 25 mmol/L glucose condition was associated with higher measured AGE levels than the 5.5 mmol/L glucose condition, particularly in the cell-associated fraction. Because no osmolality-matched control was included, these differences cannot be attributed specifically to glucose itself, and a contribution from altered extracellular osmolality cannot be excluded.

Figure 1 Advanced glycation end product levels under normal- and high-glucose culture conditions in MC3T3-E1 cells. A: Conceptual schematic showing biochemical routes by which a high-glucose environment may favor advanced glycation end product (AGE) formation and type I collagen modification. Schiff bases, Amadori products, methylglyoxal, glyoxal, and 3-deoxyglucosone may modify collagen residues and generate AGE adducts or crosslinks. This panel presents conceptual background; specific collagen adducts and crosslinks were not measured in this study; B: AGE levels in culture supernatants from MC3T3-E1 cells cultured under normal-glucose (NG) (5.5 mmol/L) or high-glucose (HG) (25 mmol/L) conditions for 3 days, 7 days, or 14 days, measured by enzyme-linked immunosorbent assay; C: AGE levels in cell lysates under the corresponding conditions; D-F: Comparison of AGE levels between culture supernatants and cell lysates on days 3, 7, and 14, respectively; G: Time-course changes in AGE levels in NG supernatants, NG lysates, HG supernatants, and HG lysates. Cultures containing 25 mmol/L glucose showed higher measured AGE levels than cultures containing 5.5 mmol/L glucose. Because no mannitol or L-glucose osmotic control was included, these differences cannot be attributed specifically to glucose metabolism, and an osmotic contribution cannot be excluded. Data are presented as the mean ± SD from three independent biological experiments (n = 3). Two-group comparisons were performed using two-tailed Welch’s t test. Time-course data were analyzed using two-way analysis of variance followed by multiple-comparison adjustment. Non-significant comparisons are not labeled. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. NS: Non-significant; AGE: Advanced glycation end product; MGO: Methylglyoxal; 3-DG: 3-Deoxyglucosone; GO: Glyoxal; FL: Fructosyllysine; CML: Nε-(carboxymethyl)lysine; CEL: Nε-(carboxyethyl)lysine; G-H1: Glyoxal-derived hydroimidazolone-1; 3-DG-H1: 3-Deoxyglucosone-derived hydroimidazolone-1; MG-H1: Methylglyoxal-derived hydroimidazolone-1; NG: Normal-glucose; HG: High-glucose.

To further explore the possible non-covalent structural compatibility of AGE-related molecules within the fibronectin-collagen peptide complex, representative AGE-related molecules, including pentosidine, Nε-(carboxymethyl)lysine, glucosepane, and MG-H1, were subjected to molecular docking using the 3GXE structure. All four molecules yielded negative AutoDock Vina scores within this model (Figure 2A and B). Among them, glucosepane showed the most negative Vina score, followed by pentosidine, MG-H1, and Nε-(carboxymethyl)lysine. Interaction analysis identified predicted hydrogen-bond, van der Waals, salt-bridge, and electrostatic contacts within the complex. These docking results were used only to describe and rank predicted non-covalent poses within this exploratory structural model and were not interpreted as evidence of COL1-specific binding, spontaneous AGE-COL1 interaction, or covalent crosslink formation.

Figure 2 Exploratory non-covalent docking within a fibronectin-collagen peptide complex and methylglyoxal-induced collagen fluorescence. A: Top-ranked docking poses of pentosidine, Nε-(carboxymethyl)lysine (CML), glucosepane, and methylglyoxal-derived hydroimidazolone-1 (MG-H1) within the crystal structure of a low-affinity collagen site in complex with fibronectin type I domains 8 and 9 (Protein Data Bank accession code: 3GXE). Each row shows the overall predicted pose, local three-dimensional contacts, and two-dimensional interaction diagram. The labeled contacts were predominantly located in fibronectin chains A and B. Specifically, pentosidine contacted aspartate (Asp) 521 and isoleucine (Ile) 519 in chain B; CML contacted tryptophan (Trp) 597, glutamine (Gln) 562, Gln560, glycine (Gly) 595, and Ile594 in chain B; glucosepane contacted Gln560, Gly593, and Ile594 in chain A and threonine 530 in chain B; and MG-H1 contacted Asp521, Ile519, and Trp553 in chain A and Gln528 in chain B. The receptor contained short collagen α1(I)-derived peptides but did not represent full-length fibrillar collagen I (COL1); B: Summary of the AutoDock Vina scores, PubChem identifiers, predicted contacts, and interaction types. More negative Vina scores were used only to rank poses within this model and were not interpreted as experimentally measured binding affinities. Each ligand underwent one docking run, and the nine generated poses were alternative solutions rather than independent replicates. Standard docking did not model glycation chemistry or covalent crosslink formation; C: Advanced glycation end product-related fluorescence in a cell-free COL1-methylglyoxal (MGO) system. Rat-tail COL1 gels (1 mg/mL; 50 μL/well) were overlaid with 100 μL phosphate-buffered saline containing 10 mmol/L MGO, with or without 20 mmol/L aminoguanidine, and incubated at 37 °C for 5 days. MGO increased COL1-associated fluorescence, whereas aminoguanidine reduced this increase. Fluorescence was measured at excitation and emission wavelengths of 370 nm and 440 nm and normalized to the COL1-only group. The MGO-only group controlled for background fluorescence. Data are presented as the mean ± SD from three reaction sets, each measured in three technical wells. Technical-well values were averaged within each reaction set. Multiple-group comparisons were performed using one-way analysis of variance followed by Tukey’s multiple-comparisons test. Ganagliflozin proline was not included in this experiment. The fluorescence assay did not identify specific glycation adducts, modified COL1 residues, or covalent crosslinks. ^dP < 0.0001. CML: Nε-(carboxymethyl)lysine; MG-H1: Methylglyoxal-derived hydroimidazolone-1; MGO: Methylglyoxal; COL1: Collagen I.

We next experimentally assessed AGE-like COL1 modification using a cell-free COL1–MGO co-incubation system. Compared with the COL1-alone group, co-incubation of COL1 with MGO markedly increased AGE-related fluorescence intensity (Figure 2C). Addition of aminoguanidine significantly reduced this fluorescence signal, whereas the MGO-alone group produced only a low background signal. These findings suggest that MGO can directly induce AGE-like modification of COL1 and that this process can be partially inhibited by an AGE formation inhibitor.

AGEs suppress MC3T3-E1 cell proliferation and define the injury model for Gana Pro intervention

To establish an appropriate model of AGE-induced osteoblast injury, MC3T3-E1 cells were treated with different concentrations of AGEs for 12 hours, 24 hours, and 48 hours. CCK-8 assays showed that low to moderate concentrations of AGEs did not markedly impair cell viability at early time points, whereas 200 μg/mL and 400 μg/mL AGEs significantly reduced cell viability, with the inhibitory effect becoming more pronounced as exposure duration increased (Figure 3A-C). Considering both injury intensity and experimental reproducibility, 200 μg/mL AGEs was selected as the main modeling concentration.

Figure 3 Concentration screening and proliferative responses to ganagliflozin proline during advanced glycation end product-modified bovine serum albumin exposure. A-C: Cell counting kit-8 metabolic-activity measurements after 12 hours, 24 hours, and 48 hours, respectively. MC3T3-E1 cells were exposed to commercial advanced glycation end product-modified bovine serum albumin (AGE-BSA) (25-400 μg/mL total protein), canagliflozin (1-40 μmol/L), or ganagliflozin proline (Gana Pro) (1-40 μmol/L). Additional groups received 200 μg/mL AGE-BSA with 1 μmol/L, 5 μmol/L, or 10 μmol/L Gana Pro. The single-agent drug series evaluated concentration-associated effects on metabolic activity and did not constitute a comprehensive safety assessment. Based on the combined concentration-screening results, 5 μmol/L Gana Pro was selected as the working concentration for subsequent experiments; it was not defined as universally optimal; D: Representative 5-ethynyl-2’-deoxyuridine (EdU) images and quantification in the normal-glucose, AGE-BSA, AGE-BSA + 1 μmol/L Gana Pro, AGE-BSA + 5 μmol/L Gana Pro, and AGE-BSA + 10 μmol/L Gana Pro groups. EdU-positive nuclei are red, and Hoechst 33342-stained nuclei are blue. AGE-BSA exposure was associated with reduced EdU incorporation, whereas 5 μmol/L Gana Pro increased the EdU-positive proportion; E: Representative propidium iodide DNA-content histograms for the corresponding groups; F: Percentages of cells in the G0/G1, S, and G2/M phases. AGE-BSA exposure was associated with increased G0/G1-phase accumulation and reduced S-phase representation. Treatment with 5 μmol/L Gana Pro partially restored S-phase representation. Data are presented as the mean ± SD from three independent biological experiments (n = 3). Technical replicate wells were averaged within each experiment and were not treated as independent observations. Multiple-group comparisons were performed using one-way analysis of variance followed by Tukey’s multiple-comparisons test. The AGE-BSA preparation lacked a matched unmodified BSA control, and its endotoxin content was not analyzed. Gana Pro-only groups were included in the metabolic-activity screen but not in the downstream molecular and osteogenic assays. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. For panels A-C, statistical comparisons were made against the corresponding normal-glucose group within each experimental series at the same time point. NG: Normal-glucose; HG: High-glucose; AGE: Advanced glycation end product; Cana: Canagliflozin; Gana Pro: Ganagliflozin proline; EdU: 5-ethynyl-2’-deoxyuridine.

The safe and effective concentration range of Gana Pro was then evaluated. Under stimulation with 200 μg/mL AGEs, Gana Pro exerted concentration-dependent effects on MC3T3-E1 cells. Gana Pro at 5 μM showed the most favorable protective effect, whereas 10 μM did not further improve cell viability and showed a potentially unfavorable trend on cell proliferation (Figure 3A-C). Therefore, 5 μM Gana Pro was selected for subsequent mechanistic and functional experiments.

The effect of Gana Pro on AGE-induced proliferative suppression was further verified using the EdU incorporation assay. Compared with the NG group, AGEs significantly reduced both the proportion of EdU-positive cells and the EdU-positive area (Figure 3D). Under AGE exposure, Gana Pro treatment restored EdU incorporation, with the most pronounced recovery observed in the 5 μM group. Consistently, flow cytometric analysis showed that, compared with the NG group, treatment with 200 μg/mL AGEs increased the proportion of MC3T3-E1 cells in the G0/G1 phase and decreased the proportion of cells in the S phase, indicating AGE-induced G0/G1-phase arrest and impaired cell-cycle progression (Figure 3E and F). Under AGE stimulation, 5 μM Gana Pro partially increased the S-phase population, suggesting that it may partly alleviate AGE-induced restriction of cell-cycle progression; however, its effect on reducing the G0/G1-phase population was not obvious. Gana Pro at 10 μM did not further improve the S-phase population and was accompanied by increased G0/G1-phase accumulation, suggesting that a higher concentration of Gana Pro may be unfavorable for cell-cycle recovery.

Gana Pro restores COL1 expression and reduces AGE-modified protein and RAGE signals

To determine whether Gana Pro altered AGE-related molecular changes, COL1, AGE-modified proteins, and RAGE were examined in MC3T3-E1 cells. Compared with the NG group, HG and AGE treatments reduced COL1 protein expression and increased AGE-modified protein immunoreactivity and RAGE expression (Figure 4A). Compared with the AGEs group, both Cana and Gana Pro increased COL1 protein expression. Gana Pro also reduced AGE-modified protein immunoreactivity and RAGE protein expression. RT-qPCR analysis showed corresponding transcriptional changes. AGE treatment reduced Col1a1 mRNA and increased Ager mRNA, whereas Gana Pro increased Col1a1 expression and reduced Ager expression (Figure 4B). These results demonstrate treatment-associated improvements in COL1 expression and AGE/RAGE-related molecular readouts.

Figure 4 Ganagliflozin proline restores collagen I expression and reduces inflammatory readouts during advanced glycation end product-modified bovine serum albumin exposure. A: Western blotting, relative-expression heatmap, and densitometric analysis of collagen I (COL1), advanced glycation end product (AGE)-modified protein immunoreactivity, and the receptor for advanced glycation end products (RAGE). MC3T3-E1 cells were cultured under normal-glucose, high-glucose, AGE-bovine serum albumin (BSA), AGE-BSA + canagliflozin (10 μmol/L), or AGE-BSA + ganagliflozin proline (Gana Pro) (5 μmol/L) conditions. Glyceraldehyde-3-phosphate dehydrogenase was the loading control. Compared with AGE-BSA alone, Gana Pro increased COL1 expression and reduced anti-AGE immunoreactivity and RAGE expression. The anti-AGE signal represents overall AGE-modified protein immunoreactivity and does not measure COL1-specific glycation or crosslinking; B: Quantitative polymerase chain reaction analysis of Col1a1 and Ager expression. Gana Pro increased Col1a1 expression and reduced Ager expression compared with AGE-BSA alone; C: Western blotting, heatmap, and densitometric analysis of nuclear factor kappa B (NF-κB) p65, phosphorylated p65 (p-p65), the p-p65/p65 ratio, and tumor necrosis factor alpha (TNF-α). Compared with AGE-BSA alone, Gana Pro reduced p-p65, the p-p65/p65 ratio, and TNF-α expression; D: Quantitative polymerase chain reaction analysis of Rela and Tnf expression. The high-glucose comparison lacked a mannitol or L-glucose osmotic control. The observed reductions in inflammatory readouts do not establish that RAGE/NF-κB signaling mediates the effects of Gana Pro. Uncropped Western blot source images are provided in the Supplementary material. Data are presented as the mean ± SD from three independent biological experiments (n = 3). Technical replicate measurements were averaged within each experiment. Multiple-group comparisons were performed using one-way analysis of variance followed by Tukey’s multiple-comparisons test. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. COL1: Collagen I; AGE: Advanced glycation end product; RAGE: Receptor for advanced glycation end products; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; NG: Normal-glucose; HG: High-glucose; Cana: Canagliflozin; Gana Pro: Ganagliflozin proline; NF-κB: Nuclear factor kappa B; p-p65: Phosphorylated p65; TNF-α: Tumor necrosis factor alpha; mRNA: Messenger RNA.

Gana Pro suppresses AGE-induced NF-κB/TNF-α inflammatory signaling

To evaluate inflammatory signaling under AGE stress, total p65, phosphorylated p65, and TNF-α were examined by Western blotting. Compared with the NG group, AGE treatment had a limited effect on total p65 expression but increased phosphorylated p65 and the phosphorylated-p65/total-p65 ratio. TNF-α protein expression also increased after AGE treatment (Figure 4C). Compared with the AGEs group, Gana Pro reduced phosphorylated p65 expression, the phosphorylated-p65/total-p65 ratio, and TNF-α protein expression. RT-qPCR analysis further showed that AGE treatment increased Rela and Tnf mRNA expression, whereas Gana Pro reduced both transcripts (Figure 4D). Thus, Gana Pro directly reduced the measured markers of NF-κB/TNF-α inflammatory signaling under AGE stress.

Gana Pro restores osteogenic marker expression and mineralization capacity under AGE stress

To determine whether Gana Pro restores osteogenic differentiation under AGE exposure, osteogenic markers were examined at both the protein and mRNA levels. Western blot analysis showed that, compared with the NG group, AGEs significantly reduced ALP, RUNX2, and OCN expression (Figure 5A). Gana Pro treatment increased the expression of these osteogenesis-related proteins, with particularly evident recovery of RUNX2 and OCN. qPCR analysis showed that AGEs downregulated Alpl, Runx2, and Bglap mRNA expression, whereas Gana Pro partially restored the transcriptional levels of these osteogenic genes (Figure 5B). RUNX2 immunofluorescence further confirmed these findings: AGE exposure weakened RUNX2 fluorescence intensity, whereas Gana Pro restored RUNX2 signal intensity (Figure 5C).

Figure 5 Ganagliflozin proline improves osteogenic and mineralization-associated readouts during advanced glycation end product-modified bovine serum albumin exposure. A: Western blotting, relative-expression heatmap, and densitometric analysis of alkaline phosphatase (ALP), runt-related transcription factor 2 (RUNX2), and osteocalcin (OCN) in advanced glycation end product-modified bovine serum albumin (AGE-BSA)-exposed MC3T3-E1 cells. Glyceraldehyde-3-phosphate dehydrogenase was the loading control. Compared with AGE-BSA alone, ganagliflozin proline (Gana Pro) increased ALP, RUNX2, and OCN expression; B: Quantitative polymerase chain reaction analysis of Alpl, Runx2, and Bglap expression. No superiority of Gana Pro over canagliflozin (Cana) is inferred unless supported by a multiplicity-adjusted direct comparison; C: Representative RUNX2 immunofluorescence images and fluorescence quantification in the normal-glucose (NG), AGE-BSA, and AGE-BSA + Gana Pro groups. RUNX2 is shown in red, and 4’,6-diamidino-2-phenylindole-stained nuclei are shown in blue; D: Representative ALP staining, ALP-positive area, and ALP activity in the NG, AGE-BSA, AGE-BSA + Cana, and AGE-BSA + Gana Pro groups. ALP-positive staining was expressed as a percentage of the analyzed field area, and ALP activity in cell lysates was expressed as U/L; E: Representative Alizarin Red S staining, positive-area quantification, and calcium content. Alizarin Red S-positive area was expressed as a percentage of the analyzed field area, whereas calcium content was reported as μg/well. Gana Pro increased both mineralization-associated readouts compared with AGE-BSA alone. Because calcium content was not normalized to protein, DNA, or viable-cell number, a contribution from differences in cell abundance cannot be excluded. Data are presented as the mean ± SD from three independent biological experiments (n = 3). Technical wells and image fields were averaged within each biological experiment and were not treated as independent observations. Multiple-group comparisons were performed using one-way analysis of variance followed by Tukey’s multiple-comparisons test. Uncropped Western blot source images are provided in the Supplementary material. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. ALP: Alkaline phosphatase; RUNX2: Runt-related transcription factor 2; OCN: Osteocalcin; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; NG: Normal-glucose; Cana: Canagliflozin; Gana Pro: Ganagliflozin proline; DAPI: 4’,6-diamidino-2-phenylindole; AGE: Advanced glycation end product; mRNA: Messenger RNA.

Osteogenic functional assays were then performed. ALP staining showed strong ALP-positive staining in the NG group, whereas AGEs markedly reduced the ALP-positive area. Consistently, ALP activity was significantly decreased after AGE treatment (Figure 5D).

Matrix mineralization was evaluated using ARS staining and calcium quantification. Compared with the NG group, AGEs markedly reduced mineralized nodule formation, the ARS-positive area, and calcium deposition (Figure 5E). Under AGE stimulation, Gana Pro increased ARS staining intensity, restored calcium content, and expanded the ARS-positive area. These results indicate that Gana Pro not only restores osteogenic marker expression but also improves early osteogenic differentiation and late-stage mineralization capacity in AGE-injured MC3T3-E1 cells.

RAGE knockdown and PMA stimulation support the functional contribution of RAGE/NF-κB-associated signaling

To assess the functional contribution of RAGE/NF-κB-associated signaling to AGE-induced injury, RAGE was knocked down using siRNA before stimulation with PMA. Compared with the NG + siNC group, the AGEs + siNC group showed increased phosphorylated p65 expression and an elevated phosphorylated-p65/total-p65 ratio (Figure 6A). RAGE knockdown reduced both responses, whereas PMA treatment increased p65 phosphorylation in AGEs + siRAGE cells. Rela mRNA showed corresponding directional changes, decreasing after RAGE knockdown and increasing after PMA stimulation (Figure 6B). These findings are consistent with RAGE contributing to NF-κB-associated signaling under AGE stress.

Figure 6 Receptor for advanced glycation end products knockdown and phorbol 12-myristate 13-acetate stimulation support pathway-associated responses during advanced glycation end product-modified bovine serum albumin exposure. MC3T3-E1 cells were assigned to normal-glucose + negative-control small interfering RNA (siNC), advanced glycation end product-modified bovine serum albumin (AGE-BSA) + siNC, AGE-BSA + receptor for advanced glycation end products (RAGE)-targeting small interfering RNA (siRAGE), or AGE-BSA + siRAGE + phorbol 12-myristate 13-acetate (PMA) (100 nmol/L; treatment duration reported in the materials and methods) groups. A: Western blotting, relative-expression heatmap, and densitometric analysis of total p65, phosphorylated p65 (p-p65), and the p-p65/p65 ratio. RAGE knockdown reduced AGE-BSA-associated p65 phosphorylation, whereas PMA increased p65 phosphorylation after RAGE knockdown; B: Quantitative polymerase chain reaction analysis of Rela expression; C: Western blotting, relative-expression heatmap, and densitometric analysis of RAGE and tumor necrosis factor alpha (TNF-α). PMA did not restore RAGE expression but increased TNF-α expression after RAGE knockdown; D: Quantitative polymerase chain reaction analysis of Ager and Tnf expression; E: Western blotting and densitometric analysis of collagen I and osteocalcin expression; F: Quantitative polymerase chain reaction analysis of Col1a1 and Bglap expression; G: Western blotting and densitometric analysis of alkaline phosphatase and runt-related transcription factor 2 expression; H: Quantitative polymerase chain reaction analysis of Alpl and Runx2 expression. RAGE knockdown increased the measured osteogenic markers, whereas PMA partially attenuated their recovery. Data are presented as the mean ± SD from three independent biological experiments (n = 3). Technical replicate measurements were averaged within each experiment. Multiple-group comparisons were performed using one-way analysis of variance followed by Tukey’s multiple-comparisons test. PMA was used as a pleiotropic stimulus capable of engaging nuclear factor kappa B (NF-κB)-associated signaling, not as a selective NF-κB agonist. Nuclear p65 translocation and nuclear/cytoplasmic fractionation were not evaluated. Ganagliflozin proline was absent from this experiment; therefore, these results do not establish that RAGE/NF-κB-associated signaling mediates its protective effects. Uncropped Western blot source images are provided in the Supplementary material. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; NF-κB: Nuclear factor kappa B; p-p65: Phosphorylated p65; NG: Normal-glucose; AGE: Advanced glycation end product; siNC: Negative-control small interfering RNA; siRAGE: Receptor for advanced glycation end products-targeting small interfering RNA; RAGE: Receptor for advanced glycation end products; TNF-α: Tumor necrosis factor alpha; mRNA: Messenger RNA; OCN: Osteocalcin; COL1: Collagen I; ALP: Alkaline phosphatase; RUNX2: Runt-related transcription factor 2.

AGE treatment also increased RAGE and TNF-α protein expression and upregulated Ager and Tnf mRNA (Figure 6C and D). RAGE knockdown reduced RAGE expression and attenuated AGE-induced TNF-α upregulation. PMA did not restore RAGE expression but increased TNF-α expression after RAGE knockdown. This response is consistent with PMA-responsive inflammatory signaling functioning downstream of RAGE.

AGE exposure reduced COL1 and OCN protein expression and decreased Col1a1 and Bglap mRNA expression (Figure 6E and F). RAGE knockdown restored these osteogenic markers, whereas PMA treatment partially attenuated their recovery. Similarly, AGE exposure reduced ALP and RUNX2 protein expression and decreased Alpl and Runx2 mRNA expression (Figure 6G and H). RAGE knockdown restored these readouts, whereas PMA treatment attenuated their recovery.

Collectively, these bidirectional pathway-perturbation results support a functional contribution of RAGE/NF-κB-associated signaling to AGE-induced inflammatory activation and osteogenic-marker suppression. Because Gana Pro was not included in this experiment, these findings do not directly establish that this pathway mediates its protective effects.

DISCUSSION

The present study identifies Gana Pro as a candidate pharmacological intervention for AGE-induced osteogenic dysfunction. Gana Pro increased EdU incorporation, restored COL1 and osteogenic-marker expression, increased ALP activity and calcium deposition, and improved matrix mineralization. These results demonstrate a protective effect across multiple molecular and functional endpoints. The improvements were accompanied by reductions in AGE-modified protein immunoreactivity, RAGE expression, p65 phosphorylation, and TNF-α expression. Separately, RAGE knockdown followed by PMA stimulation supported the functional contribution of RAGE/NF-κB-associated signaling to AGE-induced injury. Together, these datasets provide convergent mechanistic evidence for this pathway as a plausible contributor to Gana Pro-associated protection, but do not establish that pathway suppression is required for the drug response.

Diabetic bone fragility is increasingly understood as a disorder of bone quality rather than simply a consequence of reduced bone mass[2,31]. Patients with type 2 diabetes can have normal or elevated BMD despite an increased fracture risk[32,33]. This mismatch reflects contributions from altered microarchitecture, collagen matrix deterioration, abnormal AGE crosslinking, inflammation, and impaired bone remodeling[6]. Chronic hyperglycemia may therefore compromise skeletal integrity through several interacting processes that are not captured by BMD alone[2,7,18]. In the present study, higher AGE levels were observed in culture supernatants and cell lysates under 25 mmol/L glucose than under 5.5 mmol/L glucose. The relatively higher cell-associated AGE levels observed under the 25 mmol/L glucose condition suggest a locally glycative environment that may affect osteoblasts and their extracellular matrix. Because no osmolality-matched control was included, these differences cannot be attributed specifically to glucose itself.

COL1 forms the principal organic scaffold for bone apatite deposition and contributes substantially to tissue toughness[34,35]. As a long-lived matrix protein, COL1 is susceptible to cumulative non-enzymatic glycation. Experimental glycation can stiffen the bone collagen network, restrict molecular and fibrillar sliding, diminish energy dissipation, and impair fracture resistance[10]. In the present study, exploratory docking predicted possible non-covalent poses within the crystallographic fibronectin-collagen peptide complex, while MGO increased AGE-related fluorescence in a cell-free COL1 system (Figure 2). Although bulk AGE-related fluorescence has precedent in collagen-glycation studies, it does not establish covalent crosslink formation or identify specific modified residues[12,36]. Moreover, Gana Pro was not tested in the cell-free COL1-MGO system, and cellular COL1-specific glycation was not directly quantified. The current evidence therefore supports restoration of COL1 expression and reduced AGE-modified protein immunoreactivity, rather than direct inhibition of specific COL1 crosslinks by Gana Pro.

AGEs also act as bioactive stressors in osteoblastic cells, increasing oxidative stress and apoptosis while impairing proliferation, COL1 production, and ALP activity[14]. RAGE-dependent signaling contributes to AGE-induced apoptosis and mitochondrial oxidative dysfunction in osteoblastic models[18].

Previous studies have implicated AGE-RAGE, mitogen-activated protein kinase, and NF-κB-related signaling in oxidative stress, inflammation, osteoblast dysfunction, and reduced bone turnover[18,20,37]. In the present study, AGE exposure increased RAGE expression, p65 phosphorylation, and TNF-α expression. Gana Pro-mediated recovery of osteogenic outcomes occurred in parallel with reductions in these pathway readouts. These pharmacological data directly demonstrate concurrent molecular and functional improvements while leaving their mechanistic relationship unresolved.

The independent siRAGE/PMA experiment further showed that pathway manipulation altered AGE-induced inflammatory and osteogenic-marker responses. RAGE knockdown reduced p65 phosphorylation and TNF-α expression and restored COL1, ALP, RUNX2, and OCN. PMA increased p65 phosphorylation and TNF-α expression and partially attenuated osteogenic-marker recovery after RAGE knockdown. These findings accord with previous evidence of AGE-induced osteoblast dysfunction and the contribution of RAGE-associated signaling to osteoporosis[14,37]. Their convergence with the Gana Pro findings is consistent with partial involvement of the RAGE/NF-κB axis in Gana Pro-associated protection.

Nevertheless, the current design cannot determine whether reduced RAGE/NF-κB signaling mediates the protective response or partly follows broader improvement in cellular status. PMA is a pleiotropic signaling stimulus and was not used as a selective NF-κB agonist. AGE-induced injury may also involve oxidative stress, mitogen-activated protein kinase, phosphoinositide 3-kinase/protein kinase B, and Wnt/β-catenin signaling[15]. The pathway experiment should therefore be interpreted as supporting RAGE/NF-κB-associated signaling rather than demonstrating an exclusively NF-κB-specific mechanism or drug-specific pathway dependence.

Gana Pro also showed a non-linear concentration response. Among the tested concentrations, 5 μM provided the most consistent recovery of viability, EdU incorporation, and cell-cycle progression. The 10 μM concentration offered no additional benefit and was associated with greater G0/G1 accumulation. Such non-monotonic responses can arise when a compound is beneficial within a limited exposure range but less favorable at higher concentrations[38]. Accordingly, 5 μM was selected as the working concentration, although broader dose–response and exposure studies are needed before defining an optimal pharmacological range.

An osteoblast-related effect has also been reported for another SGLT2 inhibitor. Cana promoted MC3T3-E1 differentiation at 5 μM under high-glucose conditions and improved bone microarchitecture in diabetic mice, with partial involvement of adenosine monophosphate-activated protein kinase/RUNX2 signaling[29]. However, this finding was compound- and context-specific. The same study did not observe comparable skeletal effects with dapagliflozin or empagliflozin and reported inhibition of MC3T3-E1 proliferation at the highest Cana concentration tested. It therefore provides pharmacological precedent rather than evidence of a uniform class effect or of the mechanism of Gana Pro.

The broader clinical context of SGLT2 inhibition remains important. These agents have established metabolic, cardiovascular, and renal benefits in type 2 diabetes[24]. Their incorporation into diabetes and chronic kidney disease management further reflects their clinical importance[39]. However, these benefits do not establish skeletal efficacy, and bone-related findings may differ among compounds and experimental contexts. Although Cana provided a pharmacological comparator here, the present analyses do not support a claim that Gana Pro was superior without a corrected direct comparison. The data instead identify Gana Pro as a candidate intervention whose skeletal effects require validation beyond this cellular model.

The functional findings strengthen this candidate-intervention interpretation. ALP, RUNX2, and OCN represent interconnected stages of osteoblast differentiation and matrix maturation[40,41]. AGE exposure reduced these markers, ALP activity, calcium deposition, and mineralization, whereas Gana Pro improved each endpoint. The protection therefore extended beyond individual signaling proteins to functional osteogenic readouts. This distinction is relevant because diabetic bone disease ultimately manifests through impaired bone formation and matrix quality, not an isolated molecular abnormality[2].

The study also brings matrix-associated and receptor-associated AGE effects into one experimental framework. AGE-modified collagen and AGE-RAGE signaling are often discussed as complementary contributors to diabetic bone dysfunction[6]. Experimental approaches to AGE biology likewise require separation of chemical modification from receptor-mediated cellular responses[22]. Our data link high-glucose-associated AGE accumulation, COL1-related changes, inflammatory signaling, and osteogenic dysfunction within the same model. Gana Pro improved several readouts across this framework, but the data do not show that it directly engages every component. Figure 7 therefore presents an evidence-bounded working model, with drug-specific pathway mediation shown as a proposed rather than established connection. This integrated interpretation is compatible with the multifactorial nature of diabetic bone fragility[2].

Figure 7 Evidence-bounded working model of ganagliflozin proline-associated protection and receptor for advanced glycation end products/nuclear factor kappa B-associated signaling in MC3T3-E1 cells. Commercial advanced glycation end product-modified bovine serum albumin (AGE-BSA) exposure was associated with increased AGE-modified protein immunoreactivity, receptor for advanced glycation end products (RAGE) expression, p65 phosphorylation, and tumor necrosis factor alpha (TNF-α) expression. These changes occurred together with reduced proliferation, osteogenic-marker expression, alkaline phosphatase (ALP) activity, and mineralization-associated readouts. Ganagliflozin proline (Gana Pro) increased 5-ethynyl-2’-deoxyuridine incorporation, collagen I (COL1) and osteogenic-marker expression, ALP activity, calcium content, and matrix mineralization while reducing RAGE, p65 phosphorylation, and TNF-α readouts. In a separate experiment without Gana Pro, RAGE knockdown followed by phorbol 12-myristate 13-acetate (PMA) stimulation supported the functional involvement of RAGE/nuclear factor kappa B-associated signaling in AGE-BSA-associated injury. Solid connections represent relationships measured directly within the corresponding experimental module. The dashed connection between reduced pathway readouts and Gana Pro-associated functional recovery denotes a proposed contribution requiring direct Gana Pro plus PMA rescue testing. The model does not imply that Gana Pro directly inhibits COL1 glycation or crosslinking. Because matched unmodified bovine serum albumin and endotoxin-controlled conditions were absent, the initiating injury cannot be attributed exclusively to AGE modifications. The model is restricted to MC3T3-E1 cells and does not imply validation in primary, human, or in vivo osteogenic systems. AGE: Advanced glycation end product; OCN: Osteocalcin; COL1: Collagen I; ALP: Alkaline phosphatase; RUNX2: Runt-related transcription factor 2; TNF-α: Tumor necrosis factor alpha; NF-κB: Nuclear factor kappa B; RAGE: Receptor for advanced glycation end products; p-p65: Phosphorylated p65; Gana Pro: Ganagliflozin proline.

Several limitations should be acknowledged. First, the study used only MC3T3-E1 cells. Validation in primary osteoblasts, bone marrow mesenchymal stem cells, human osteoblasts, and in vivo diabetic bone models is required[6]. Second, the high-glucose experiment lacked an osmotic control, preventing complete separation of glucose-specific effects from altered osmolality. In addition, the commercial AGE-BSA preparation lacked a matched unmodified BSA control and endotoxin assessment. Therefore, the observed cellular and inflammatory responses should be interpreted as responses occurring during commercial AGE-BSA exposure rather than as definitive AGE-specific effects. Third, the absence of a Gana Pro-only group limits assessment of its baseline effects. Finally, ALP activity was expressed as U/L of cell lysate, and calcium content was reported as μg/well. These measurements were not normalized to total protein, DNA content, or viable cell number; therefore, differences in cell abundance may have contributed to the observed group differences.

The analytical evidence for COL1 modification also remains indirect. Docking predicts non-covalent structural compatibility, while AGE-related fluorescence does not identify individual glycation products or COL1-specific adducts[42]. Future work should use mass spectrometry and collagen-crosslink profiling to quantify glucosepane, pentosidine, and other modifications directly. Second-harmonic generation imaging and biomechanical testing could then determine whether these chemical changes alter fibrillar organization and material properties. Other AGE-responsive processes, including mitochondrial injury, autophagy, ferroptosis, and oxidative stress, may also influence osteogenic function[43,44]. Their relationship to the Gana Pro response remains to be determined.

An important mechanistic limitation is that the pathway-perturbation experiment did not include an AGEs + Gana Pro + NF-κB-activating stimulus group. Consequently, this study cannot determine directly whether forced reactivation of NF-κB-associated signaling attenuates the protective effects of Gana Pro. A priority experiment should include NG + vehicle, NG + PMA, AGEs + vehicle, AGEs + Gana Pro, AGEs + PMA, and AGEs + Gana Pro + PMA groups. The phosphorylated-p65/total-p65 ratio and ALP activity should serve as the primary signaling and functional endpoints, respectively. RUNX2, ALP, and OCN expression would provide complementary outcomes. Within the AGE-exposed groups, the primary analysis should test the Gana Pro × PMA interaction. Only if PMA restores p65 phosphorylation and significantly attenuates Gana Pro-associated functional benefits would the findings support a partial mediating contribution of NF-κB-associated signaling. Even then, the pleiotropic actions of PMA would preclude attribution to NF-κB alone.

Finally, translational development will require model-informed characterization of exposure-response relationships[45]. Direct evaluation of pharmacokinetics, pharmacodynamics, bone distribution, dose dependence, tolerability, and long-term safety will also be necessary[46]. Until such evidence becomes available, the findings should be interpreted as evidence of a protective cellular phenotype and support for a plausible pathway contribution, not clinical anti-osteoporotic efficacy.

CONCLUSION

Gana Pro attenuated impairments in osteoblast proliferation, differentiation, and matrix mineralization observed during commercial AGE-BSA exposure and restored COL1 expression. These protective effects were accompanied by reductions in AGE-modified protein immunoreactivity and RAGE/NF-κB/TNF-α-associated inflammatory readouts. Independent RAGE-knockdown and PMA-stimulation experiments supported the functional contribution of this pathway to osteogenic suppression observed during AGE-BSA exposure. The convergence of these findings identifies RAGE/NF-κB signaling as a plausible mechanistic contributor to Gana Pro-associated protection, while direct drug-specific pathway dependence remains to be established. Gana Pro therefore warrants further preclinical evaluation as a candidate intervention for osteoblast dysfunction associated with commercial AGE-BSA exposure.

ACKNOWLEDGEMENTS

Gana Pro was provided as an in-kind research sample by Jilin Huikang Pharmaceutical Co., Ltd. and Hangzhou Zhongmei Huadong Pharmaceutical Co., Ltd. (Hangzhou, Zhejiang Province, China). Zhejiang Luming Biotechnology Co., Ltd. (Wenzhou, Zhejiang Province, China) provided technical assistance with selected experimental procedures. Neither company participated in the study design, data interpretation, manuscript preparation, or decision to submit the manuscript. The authors retained full control over the data and publication.