Abstract
Adult articular cartilage chondrocytes have a limited capacity to divide compared to juvenile cells, but the mechanisms behind this decline remain unclear. This study investigates metabolic changes associated with the cessation of chondrocyte proliferation in mouse articular cartilage. Using 5–ethynyl–2′–deoxyuridine (EdU) labeling, the postnatal decline in proliferation was tracked. Label-free fluorescence-lifetime imaging microscopy (FLIM) method, combined with artificial intelligence (AI)-assisted image segmentation, was applied to live cartilage sections to analyze metabolic parameters. Results showed that 1-month-old articular cartilage chondrocytes enter quiescence with significant changes in FLIM fluorescence decay parameters across cartilage zones compared to juvenile chondrocytes. Chondroprogenitors in the superficial zone showed a gradual decrease in citrate synthase content, while glycolytic activity increased with tissue depth. These findings reveal metabolic reprogramming that enables chondrocytes to adapt their metabolism despite limited oxygen availability to meet functional demands. Investigating these chondrocyte adaptations provides key insights for identifying metabolic targets and improving the design of durable, well-integrated tissue-engineered cartilage.
Keywords
Introduction
Synovial joint formation commences in the unbroken cartilaginous anlagen (ANL) of developing limbs during embryogenesis. 1 Postnatal synovial joint morphogenesis proceeds due to the epiphyseal cartilage. 1 The articular cartilage and the growth plate cartilage are demarcated by a secondary ossification center (SOC) formed in the early postnatal period. 2 Articular cartilage serves to reduce friction of bone epiphyses in a joint and to absorb impact forces associated with mobility, and growth plate plays a pivotal role in promoting longitudinal bone growth.1,3 The articular cartilage and growth plate tissues are both hyaline cartilage, however, they have different functionality and cell arrangement in the structure. In the early postnatal period, cells of the articular cartilage and growth plate proliferate actively.4,5 Once joint formation is complete, chondrocytes cease dividing, and growth plate fusion is finalized (approximately at 20 years of age in humans 6 ); however, articular cartilage assumes a permanent state that persists throughout the lifespan. Therefore, the growth plate and the articular cartilage are so-called “transient” and “permanent” hyaline cartilages respectively.
Due to its avascular nature, the hyaline cartilage cells obtain glucose and oxygen from the synovial fluid and have the ability to adjust to low-oxygen conditions. 7 Oxygen supply to cartilage from subchondral bone remains restricted, occurring mainly via defects or channels in the subchondral plate; however, this pathway is not predominant in healthy cartilage. In vitro measurements supported the view that, in cartilage, adenosine 5′-triphosphate (ATP) is generated predominantly by the glycolytic pathway. 8 Anaerobic glycolysis and mitochondrial OXPHOS are well-known pathways for producing ATP. One molecule of glucose could generate more than 30 molecules of ATP in OXPHOS.9,10 Compared with the number of molecules obtained by mitochondrial respiration, glycolytic activity is rapid, and a small number of ATP molecules are generated at a very fast rate.11,12 The glycolysis production of lactate from glucose occurs 10–100 times faster per unit of glucose. 12 In fact, the amount of ATP produced in the same period of time through glycolysis and mitochondrial respiration is comparable. 12 The most important advantage of glycolysis generating ATP is the possibility to use glycolytic intermediates (e.g. nucleotides, amino acids, and lipids) as tissue building elements.13,14
Articular cartilage cells are known to be capable of active proliferation at a young age but do not produce new chondrocytes in adulthood.4,5 Animal models, such as mice, are advantageous tools in the research of articular cartilage development. 15 Studies on mice in the past decade have shown that the articular cartilage contains a progenitor cell population in superficial zone of articular cartilage, which serve as progenitors giving rise to chondrocytes throughout the entire articular cartilage by appositional growth postnatally.1,5,16,17 It was observed that the renewal of articular cartilage chondrocytes is retarded in adult mice after postnatal day 30 (P30) considerably, 5 while both before and after P30, the chondrocytes display chondrogenic markers, and the articular cartilage tissue maintains the zonal structure.
Proliferating cells are known to have important metabolic requirements that extend beyond ATP generation. 14 To produce two viable daughter cells at mitosis, proliferating cells preferentially generate biomass from glucose molecules using glycolytic intermediates (e.g. nucleotides, amino acids, and lipids) as building elements for new cells.14,18 The proliferating cells are generally accepted to require close proximity to blood vessels for prompt access to oxygen and nutrients. However, chondrocytes have a relative deficiency of access to oxygen and glucose, as well as limited pathways for metabolite removal, 19 at the same time they are capable of changing their proliferation state. It is widely known that certain types of actively proliferating cells, such as cancer cells, prefer glycolysis even when there is an abundance of oxygen (the “Warburg effect” or “aerobic glycolysis”). 20
The mechanisms underlying the transition of chondrocytes in articular cartilage from a proliferative to a quiescent state, along with associated metabolic changes during joint maturation, remain unclear. This study aims to identify the metabolic states accompanying chondrocyte maturation and the shift from proliferation to quiescence—key factors that limit the regenerative capacity of adult cartilage. Elucidating how metabolic shifts correspond to chondrocyte proliferative and differentiative states may reveal opportunities to improve the quality of engineered chondrogenic tissues via targeted metabolic modulation. Unlike most metabolic investigations,21,22 which are performed in cell cultures, this study was conducted on living mouse tissue sections using label-free fluorescence lifetime imaging microscopy (FLIM).
FLIM, noninvasive optical technique, allowed us to investigate the cellular metabolic alterations on living tissue sections avoiding cartilage matrix destruction and substantially preserving chondrocyte native state and their milieu. Nicotinamide adenine dinucleotide reduced form (NADH), a crucial intracellular coenzyme in cellular energy metabolism, exhibits autofluorescence, which decay parameters (FDPs) are highly sensitive to the cell’s metabolic state. By analyzing the fluorescence lifetimes (τ1 and τ2) and corresponding amplitudes (a1 and a2) of the short- and long-lived components of NADH autofluorescence, FLIM enables quantitative, intensity-independent assessment of the relative concentrations of free versus protein-bound NADH. This, in turn, provides a robust proxy for evaluating the cellular redox state. 23 Free NADH is characterized by a short fluorescence lifetime (τ1) of ~400 ps, whereas protein-bound NADH exhibits longer lifetimes (τ2), typically ranging from 1.0 to 4.0 ns, depending on the specific binding protein. Because the absolute level of protein-bound NADH tends to remain relatively constant, the ratio of free to bound NADH reflects the NAD⁺/NADH redox balance, which is also captured by the mean fluorescence lifetime (τm). Notably, τm is additionally influenced by the specific binding partners of NADH, and thus can vary independently of the free-to-bound NADH ratio. In contrast, the amplitude ratio (a1/a2), which represents the relative fluorescence contributions of free and bound NADH, is less affected by differences in protein composition. 23 By employing the FLIM technique with AI-enhanced image segmentation, the study aimed to (1) assess general trends in metabolic status alterations and (2) examine these changes in relation to the spatial organization of cells within the superficial, transitional, and deep zones of cartilage.
Methods
Animals
Animals were housed in accordance with the Guidelines for the care and use of laboratory animals (GOST 33216-2014, Russia) based on “European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes” (Strasbourg, 1986.03.18). Mice C57Bl/6 Black, an inbred strain without genetic modifications, were housed in a Specific pathogen-free (SPF) animal research facility under controlled conditions (24°C, 12-h light/dark cycle, 50% humidity) and routinely provided with food and water, as well as shelters and nest boxes. The Committee on Animal Research and Ethics at the I.M. Sechenov First Moscow State Medical University (Sechenov University) approved all mouse experiments (22-01-e of 29 December 2022, 22-03-e and 22-04-e of 28 February 2023). Healthy littermate mice of the selected ages, with no visible abnormalities in appearance or behavior, were included in the study. Since the experimental groups included mice in the early postnatal period, no sex-based differentiation was applied; both male and female animals were included in the study. The experimental animals were monitored 2–3 times per week to control their health status. The study design and sample sizes were selected with a commitment to the concepts of the Three Rs. 24 The work has been reported in line with the ARRIVE guidelines 2.0.
Tissue collection
The animals were euthanized by cervical dislocation following deep anesthesia. Harvested hind limbs were fixed in 10% formalin (ITW Reagents, Italy) along 20–24 h (for the thymidine analog 5–ethynyl–2′–deoxyuridine (EdU) detection) and 6 h (for immunofluorescence staining) at +4°C, decalcified into 10% EDTA (pH 8.1) water solution for 3 days, kept in 30% sucrose in PBS at +4°C for 2 h, embedded into cryopreservation compound (4583, Sakura Finetek, Japan) and frozen at −80°C. By using a cryostat (HM525 NX UV, Thermo Fisher Scientific, USA), the frozen blocks were cut into 30 μm sections.
EdU-labeling, axitinib injections and fluorescence detection
EdU (5–ethynyl–2′–deoxyuridine, thymidine analog; 40540; Lumiprobe, USA) was dissolved in sterile saline and, for a cumulative effect, injected intraperitoneally (52 μg/g) four times in 3-day periods before mice were sacrificed. The time points of EdU-labeled cell detection were selected before the start of the SOC formation within the epiphysis of a knee femur (P5 and P14), during a completion of SOC formation (P20 and P31) and in the adult stage (P65) (scheme at Figure 1(B)). To collect all data, one field of view per section, three tissue sections per mouse, one joint per mouse, and three animals per selected age group were included in the experiment (n = 15 animals in total).

Changes in the number of proliferating cells in articular cartilage during postnatal joint development. (A) Representative images depicting distinct stages of postnatal development of murine femoral articular cartilage (4× objective). The images were rotated; the background removal was performed using the Background Eraser Tool in Adobe Photoshop CS6 v. 13.0.1. (B) Scheme for intraperitoneal injections of EdU and Axitinib in mice. (C) Joint 30 μm sections of mouse tissue after injections of EdU; (Ca–Ce) Safranin O/Fast Green staining (20× objective); (Ca′–Ce′) LSM confocal images for EdU detection (green) with a transmitted light (gray) (10× objective); (Ca''–Ce'') LSM confocal images for EdU detection (green) with DAPI (blue) (40× objective). (D) Joint 30-μm sections of mouse tissue after injections of EdU and Axitinib; (Da-De) Safranin O/Fast Green staining (20× objective); (Da'–De') LSM confocal images for EdU detection (green) with a transmitted light (gray) (10× objective); (Da″–De'″) LSM confocal images for EdU detection (green) with DAPI (blue) (40× objective). The yellow dotted lines in panels (C) and (D) indicate the boundaries of the articular cartilage zones. (E) Comparing EdU-labeled cell numbers from different age groups, including mice injected with Axitinib. Data were collected from one field of view per section, three tissue sections of one joint per mouse, and three different mice (n = 3) per group.
EdU incorporated into de novo–synthesized genomic DNA was revealed by a click reaction on cryosections covered by the mixture of 1 M Tris (pH 7.5), 100 mM CuSO4, 0.5 M L-ascorbic acid, and 2 mM Azide-fluor-488 (760765, Sigma) for 30 min at room temperature (RT).
Axitinib (PZ0193-5MG, Sigma) is a second generation tyrosine kinase inhibitor that selectively inhibits vascular endothelial growth factor receptors (VEGFRs), 25 it was dissolved in DMSO (10 mg/ml; 131954.1611, PanReac_AppliChem) and injected intraperitoneally (15 μg/g) daily from P10 to P13 before the EdU injections (Figure 1B). Prolonging of axitinib injections was lethal. Three animals of each age were included in the experiment (n = 9 animals in total).
Three-dimensional representations of fluorescence signal detection based on a set of tissue optical sections across the z-axis (Z-stack images, 1.3 μm step) were obtained using the confocal laser scanning microscope (LSM) FluoView FV3000 (Olympus, Japan). EdU-labeled cell counting was accomplished in 80x260 µm rectangular area of the maximum intensity projection of Z-stack images of 30 μm tissue sections, including the superficial (SF), transitional (TR; middle), and deep (DP) zones of articular cartilage using the open-source platform for biological-image analysis Fiji/ImageJ 26 with StarDist_2D plugin. 27
Harvesting of osteochondral explants, treatment with axitinib and EdU
At postnatal day 6 (P6), fresh, viable osteochondral explants were collected from the femoral epiphyses of both knee joints to a 4°C phosphate buffer immediately after euthanizing mice (n = 4). The explants were transferred to a 24-well plate containing chondrogenic medium with high-glucose (4.5 g/L) DMEM (G4517, Servicebio, China), supplemented with 2 mM L-glutamine (Paneco-Ltd., Russia), 10−4 M 2-mercaptoethanol (Thermo Fisher Scientific, USA), 5 µg/ml ascorbic acid, Insulin-Transferrin-Selenium (Bioinlabs, Russia), 0.5% bovine serum albumin (Servicebio, China), 0.35 mM L-proline (Paneco-Ltd., Russia), Penicillin-Streptomycin (10,000 U/mL) solution (Gibco, USA), 10−7 M dexamethasone (Sigma, USA), and 10 ng/ml TGF-β3 (Yeasen Biotechnology (Shanghai) Co., Ltd., China). Axitinib dissolved in DMSO (ITW Reagents, Italy) was added to achieve a final concentration of 10 µM, while control samples received DMSO alone. Explants were cultured in a 5% CO2 incubator at 37°C. The culture medium was changed every day for 4 days prior to EdU treatment (scheme at Supplemental Figure S2A). EdU (20 µM) was added to the chondrogenic medium containing axitinib and incubated for 65 h. The explants were subsequently fixed in 10 % formalin at +4°C for 6 h, kept in 30% sucrose in PBS at +4°C for 1 h, embedded in a cryopreservation compound, sectioned (30 μm thickness) and then processed for immunostaining and click chemistry labeling of EdU. One field of view per section, three sections per explant, and four explants per group were analyzed.
Isolation of primary chondrocytes, treatment with axitinib and EdU
Primary chondrocytes were isolated from the knee cartilage of P6 mice (n = 3) through enzymatic digestion using 0.1% collagenase II (Worthington Biochemical Corp., NJ, USA) at 37°C with gentle agitation for 15 h. The cells were passaged as a 2D culture onto 24 mm square glass coverslips (Epredia, USA) in DMEM/F12 medium (Gibco, USA) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, USA) and 10 µg/ml gentamicin (Paneco-Ltd., Russia) at a density of 2 × 105 cells per coverslip. Axitinib dissolved in DMSO (ITW Reagents, Italy) was added to the culture medium to achieve a final concentration of 10 µM, while control samples received DMSO alone. The cells were cultured in a 5% CO2 incubator at 37ºC, and the culture medium was changed every day for 4 days prior to EdU treatment (scheme at Supplemental Figure S2A). EdU (20 µM) was added to the culture medium containing axitinib and incubated for 65 h. Coverslips with cells were fixed in 3.7% PFA for 15 min and then processed for immunostaining and click chemistry labeling of EdU.
Histology and immunofluorescence protocols
The 30 μm sections were stained for Safranin_O/Fast Green (477736, 2353459; Merck, Germany) according to standard staining protocol. 28 For immunofluorescence staining, the antibodies against Lactate Dehydrogenase A (Ldha) (ab300637; Abcam, Cambridge, UK; dilution 1:200), Citrate synthase (Cs) (MA5-17264, Invitrogen, USA; dilution 1:200), Cd73 (551123; BD Pharmingen™, USA; dilution 1:100), Myocyte-Specific Enhancer Factor 2C (Mef2c) (HPA005533; Sigma; dilution 1:200), and Sox9 (HPA001758; Sigma; dilution 1:200) were used. In addition, an antibody recognizing endogenous levels of Pyruvate dehydrogenase α1 protein only when phosphorylated at Ser293 residue (pPdh) was applied (31866; Cell Signaling Technology, USA; dilution 1:50). For metabolic marker antibodies, antigen retrieval protocol included heating samples in citrate buffer (pH 6.0) at 95°C for 5–20 min. For antibodies against Cd73 and Sox9, tissue sections were incubated in citrate buffer (pH 6.0) at 95°C for 7 min followed by incubation with 0.1% Trypsin in PBST (PBS + 0.1% Tween-20) at 37°C for 10 min. Background signal was blocked using 3% horse serum (16050130, Gibco™, NY, USA) combined with 0.1% Triton X-100 (ITW Reagents, Italy) for permeabilization, incubated for 1 h at room temperature (RT). Sections as well as coverslips with cells were incubated with primary antibodies overnight at +4°C. Fluorescent secondary antibodies (dilution 1:500) were applied afterward for 1.5 h at RT in the dark. Alexa Fluor™ 647 goat anti-mouse IgG (115-605-003, Jackson ImmunoResearch Europe Ltd., UK) and Alexa Fluor™ 488 donkey anti-rabbit IgG (A-21206, Invitrogen, USA) were used for Cs and Ldha co-staining. Cy™3 AffiniPure® donkey anti-rabbit IgG (711-165-152, Jackson ImmunoResearch Europe Ltd., UK) were used to stain Sox9 or pPdh. Cyanine3 goat anti-mouse IgG (A10521, Invitrogen, USA) and Alexa Fluor™ 647 goat anti-rabbit IgG (A-21245, Invitrogen, USA) were used to stain Cd73 and Mef2c, respectively. Negative control did not include primary antibodies. DAPI (4′,6-diamidino-2-phenylindole) (TC229; HiMedia Laboratories, Mumbai, India) nuclear staining (stock 10 mg/ml; dilution 1:500) was applied for 15 min at RT in the dark. The fluorescence was detected by the confocal laser scanning microscope (LSM) FluoView FV3000 and FV31S-DT software (Ver.2.6) (Olympus, Japan). To collect data for analysis of metabolic enzymes fluorescence, one field of view per section, three tissue sections per mouse, one joint per mouse, and three different mice (n = 3) per age group were used.
Nitrotetrazolium blue staining for assessing ROS content in living cartilage tissues
Distal ends of both femurs and femoral heads were harvested from P14 and P31 mice (n = 5 animals per group) and collected in Hanks’ balanced salt solution (HBSS) maintained at 4°C. Subsequently, the samples were incubated with 0.2% w/v nitrotetrazolium blue chloride (NBT) (CAS 298-83-9, Sigma-Aldrich, USA) dissolved in HBSS at 37°C for 40 min under gentle agitation in a CO2 incubator. For the positive control, femoral epiphyses were initially incubated with 2 mM menadione sodium bisulfite (Vikasol, OJSC “Dalkhimpharm,” Russia) in HBSS at 37°C for 30 min with gentle agitation in a CO2 incubator. Femoral heads were utilized as negative controls and incubated in HBSS. Subsequently, the samples underwent fixation, decalcification, and cryopreservation, as previously described (subsection 2.2). Tissue sections with a thickness of 20 μm were imaged using a light microscope (DM1000 LED, Leica, Germany) equipped with SC50 camera and cellSens software (Olympus Corporation, Japan). One tissue section per mouse and five mice per age were analyzed.
Quantification of fluorescence intensity
Quantification of fluorescence intensity for Ldha, pPdh, and Cs using the Z-axis projections of confocal LSM images (1.3 μm step along Z-axis) was conducted in Fiji/ImageJ, an open source software. 26 Fluorescence intensity was measured as the mean gray value of the individual cell. Background subtraction was performed for each value. All data were collected from one field of view per section, three tissue sections per mouse, one joint per mouse, and three mice per group.
Fluorescence-lifetime imaging microscopy (FLIM)
Tissue sections preparation
To assess the metabolic state of articular cartilage cells in mice at P14 and P31 of joint development, two-photon excited FLIM with endogenous contrast was performed. FLIM is a tool that can measure the ratio of free and bound NADH cofactors to detect changes in cell metabolism. 23 Within the intrinsic fluorophores, NADH also has a good (0.1–0.25 GM) two-photon cross section around 720 nm, 29 allowing two-photon excitation (2PE) and therefore achievement of a good penetration depth for imaging in tissue. 23 Two knee joints from each P14 and P31 mouse (n = 5 littermate animals per group) were harvested and collected in cold PBS. The living joint tissues were manually sectioned (~500 μm) on ice using a razor blade. To prevent sample drying and heating, FLIM measurements were performed in cold sterile PBS with a small cooling agent pressing the sections down to the glass bottom of a confocal dish (FluoroDish™, WPI, China). FLIM measurements were performed on living femoral articular cartilage and were started in 20 min post-mortem. Total of 37 and 27 FLIM images was analyzed in the P14 and P31 groups, respectively.
Imaging parameters
Two-photon FLIM was performed using the confocal laser-scanning microscope LSM 880 (Carl Zeiss, Oberkochen, Germany). System was equipped with a short-pulse femtosecond Ti:Sa laser Mai Tai HP (Spectra-Physics, Milpitas, CA, USA) with a pulse repetition rate of 80 MHz and a pulse duration of 140 fs used for excitation and a FLIM module Simple Tau 152 TCSPC (Becker and Hickl GmbH, Berlin, Germany) providing temporal resolution of ~10 ps in 20 ns range for fluorescence decay detection. The Objective C Plan-Apochromat 40×/1.3 NA Oil DIC M27 was used for image acquisition. Two-photon-excited fluorescence at a wavelength of 750 nm (average power of ~6 mW on the sample) was detected in the 450–500 nm range (optimal excitation and detection parameters of NADH). FLIM-stacks with size of 512 × 512 pixels (field of view of size 212.5 × 212.5 µm) were captured with an acquisition time of 120 s, to assure the minimum number of photons per pixel of the cells to be ~5000.
AI-assisted segmentation of FLIM data with segment anything model
The analysis of FDPs of individual cartilage cells with respect to their location in the structure of the cartilage tissue included three main steps (Supplemental Figure S1A): cell segmentation, zoning of cartilage tissue, and extraction of the cells’ FDPs with its linking to the appropriate group of data. The cell segmentation was performed using fluorescence intensity images before calculating fluorescence decay (Supplemental Figure S1Aa). The fluorescence intensity images were normalized to [0,255] intensity range and then fed into the segmentation model, which assigned each pixel of the input image to the object of interest (cell) or background. For segmentation, an original approach based on the use of a state-of-the-art zero-shot Segment Anything Model (SAM) with pre-trained weights was used, 30 which predicted 512 × 512 binary segmentation maps, bounding boxes and labels for the unique regions of the input fluorescence intensity images (Supplemental Figure S1Ab).
A segmentation model from the SAM family was used to delineate regions corresponding to individual cells in fluorescence images (FLIM files) (Supplemental Figure S1). These models are designed to identify objects within an image (in our case, individual cells) for subsequent quantitative analysis. 30 At its core, the SAM model is an artificial intelligence (AI) system pre-trained to recognize the regions and boundaries of objects in images—a task known as image segmentation. Specifically, the AI identified regions (cells) on FLIM images to determine cellular coordinates (distances to the cartilage tissue surface) and to calculate fluorescence decay parameters (FDPs) separately within each delineated cell. The advantage of using such a “foundation model” is that it performs the assigned image segmentation task accurately without requiring manual labeling of thousands of cells for training. It also does so consistently across a large number of images, eliminating inter-observer variability and greatly accelerating the analysis process. We hypothesized that the SAM pre-trained model would generalize well to our specific fluorescence microscopy images of cartilage without the need for additional training or fine-tuning for tissue type.
The SAM model successfully segmented >90% of the cells with relatively high fluorescence intensity in the images. The 64 FLIM images were analyzed and a total of 5480 objects were segmented. To ensure accurate segmentation of cartilage cells in the images, the results were manually inspected, and regions incorrectly identified as cells (Supplemental Figure S1Ac) were discarded, reducing the number of detected objects to 3889 (3201 in the P14 group and 688 in the P31 group).
To link cells’ location in the structure of cartilage, their morphology and FDPs, two procedures were used: (1) manual selection of zones of cells similar by morphology and distance from the cartilage surface (Supplemental Figure S1Ad) and (2) automatic calculation of the distance from the center of each cell to the surface of the articular cartilage (Supplemental Figure S1Ae–Af). In the first approach, three cell zones of articular cartilage were determined: superficial (SF) zone with flattened chondroprogenitors, transitional (TR) zone with round chondrocytes, and deep (DP) zone with larger cells (Supplemental Figure S1Ad). As an alternative, distances from the center of each segmented cell to the nearest cartilage surface point in the images were calculated (Supplemental Figure S1Ae–Af). The mean values of the FDPs belonging to individual segmented regions were calculated and used to represent individual cells in the subsequent analysis. To extract FDPs, fluorescence decay curves in each pixel of an image were fitted using biexponential decay model convoluted with the instrument response function. The fluorescence signal in each pixel was preliminary binned with spatial binning of four, that is, fluorescence decays were averaged in a sliding window of 9 × 9 pixels. The parameters were chosen to match the excitation and emission maxima of NADH fluorescence. Since the fluorescence lifetime of free form of NADH is weakly dependent on the environment, 31 the decay lifetime of the fast component τ1 was fixed at 0.4 ns. As independent parameters, the decay time of the long-living component τ2 and the amplitudes of the fast and long-living components a1 and a2 were reconstructed, and the average fluorescence decay time calculated as τm= (a1τ1 + a2τ2) /(a1 + a2) (Supplemental Figure S1Ag–Ai).
FLIM data fitting was performed using SPCImage 8.6 software, while its subsequent analysis—cells segmentation and data aggregation were performed using custom-written Python scripts using PyTorch, 32 SciPy, 33 Matplotlib, 34 and Scikit-Image libraries. 35
Live/dead staining protocol
Immediately after FLIM measurements, tissue sections were stained according to the Live/Dead staining protocol to confirm the cell viability in tissue sections (Supplemental Figure S1B). Briefly, the sections were placed into 10 μM Calcein_AM solution (C3100MP; Invitrogen; USA) for 30 min at RT in a dark box followed by incubation in 0.3 μM propidium iodide (P4170-10MG, Sigma-Aldrich, Germany) solution for 5 min at RT. The samples were examined using the LSM 880 (Zeiss, Germany). Fluorescence images were processed using microscope software ZEN 2.0 (black edition) (Zeiss, Germany). Data were collected from one field of view per joint, two joints per mouse; n = 5 different mice per age.
Osteochondral injury model
The animals were randomly assigned to three groups (osteochondral defect at P31, osteochondral defect at P14, sham surgery at P31). Three animals were included in each experimental group (n = 9 animals in total). Focal osteochondral defects with 300 μm diameters were made in trochlear groove of distal femoral epiphysis of mice knee using a custom electric drill. Mice were anesthetized using intraperitoneal injections of Zoletil® 100 (tiletamine-zolazepam, 10 μg/g, Virbac, France) and Rometar (xylazine hydrochloride, 10 μg/g, Bioveta a.s., Czech Republic), nonnarcotic, nonbarbiturate, injectable anesthetic agents. Mice were checked for the absence of reactionary movements as a sign of sufficient anesthesia. Corneregel eye ointment (Dr. Gerhard Mann Chem.-Pharm. Fabrik, GmbH, Germany) was applied to prevent eye drying. Knee surgery was made via a long lateral para-patella skin incision, synovial capsule was opened, and the patella was dislocated and flip laterally to expose the trochlear groove. The incision was stitched with a rapidly-absorbing suture (VICRYL™ 6.0, Ethicon (Johnson&Johnson), USA). The animals were then placed on a heating pad to maintain core body temperature at 37°C until the first signs of awakening. Mice showed weight-bearing activity within 2 h after surgery. After the surgeries, the animals were monitored every 12 h for the first 3 days to detect their mobility and identify any signs of distress. The study of osteochondral injury was conducted with a support of the state assignment of the Ministry of Health of the Russian Federation (Theme No. NZAF-2024-0016).
Statistical analysis
Statistical analyses of FLIM data and fluorescence measurements of metabolic enzymes were performed using R (version 4.4.2, The R Foundation for Statistical Computing, Vienna, Austria) in RStudio (version 2025.05.1, Posit Software PBC, USA). A linear model or linear mixed-effects model (LMM) was used to evaluate the effects. R package “lme4” 36 was applied to account for hierarchical structure of the data and multiple measurements within individual animals. Significance was tested using type III ANOVA with Satterthwaite approximation, followed by Tukey-adjusted post-hoc pairwise comparisons using estimated marginal means (“emmeans” R package 37 ).
EdU-labeled cell counting was analyzed in the statistical software platform SPSS® Statistics 17.0 (IBM®, USA). To calculate p-values, multiple treatments were analyzed by factorial ANOVA followed by a Bonferroni post-hoc test to perform the pairwise comparisons between groups.
Animal experiments were performed with age-matched mice. A p-value ⩽ 0.05 was considered significant. Box-and-whisker plots were generated using GraphPad Prism version 8.0.1 (Dotmatics, Boston, USA).
Results
The number of EdU-labeled cells in the mouse articular cartilage was decreasing during the first postnatal month
Replication of genetic material is a key process underlying cell division. The EdU-labeling method was made to tag replicating DNA allowing the nuclei of dividing cells and their progenies to be marked due to ability of the tag to remain within the replicated DNA for prolonged periods. 38 The method is based on the incorporation of EdU, synthetic nucleoside analog of thymidine, into replicating DNA and its subsequent chemical detection, frequently called a “click” reaction. 39
To find the optimal time points for FLIM measurements, the cell division dynamics of articular cartilage in the first month after birth were studied, since the active renewal of cartilage tissue in mice was previously known to occur before postnatal day 30 (P30). 5 To detect and quantify actively proliferating cells in the articular cartilage, juvenile mice received intraperitoneal injections of EdU incorporated into de novo–synthesized genomic DNA 40 (Figure 1). The key changes in the structure of the developing joint during the postnatal period guided us to select the appropriate time points for EdU injections. The postnatal day 5 (P5) was selected as the time point before the formation of the secondary ossification center (SOC) and before articular and growth plate cartilage dividing within the epiphysis of a knee femur (Figure 1(A)). The SOC formation process began on postnatal day 14 (P14) (Figure 1(A)). The selection of two more points, P20 and P31, coincided with the development and completion of the SOC formation (Figure 1(A)). Point P65 was chosen to observe the fully formed mouse’s knee joint during the adult stage. For cumulative effects, four EdU injections were administered within 3 days before the animals were sacrificed at P5, P14, P20, P31, and P65 (scheme at Figure 1(B)).
Counting the EdU-labeled cells found in both superficial and transitional zones of articular cartilage showed a gradual decline in the number of proliferating cells until day P65 (Figure 1(C)–(E)). Due to their morphology, hypertrophic chondrocytes surrounding SOC were easily visible in sections (especially at P14 and P20), and they were not labeled with EdU (Figure 1(Cb-b″) and (Cc-c″)), which was correlated with the results of research on growth plate cartilage and in vitro studies of chondrocytes. 41 In the articular cartilage of the P65 mouse group, individual EdU-labeled/proliferating cells were observed very rarely (Figure 1(Ce′–e″)). EdU-positive chondrocytes were counted within a fixed-area rectangle consistently positioned at the similar region of the femoral section. In our case, the fixed-area rectangles within most cartilage sections from P65 mice did not include any EdU-positive cells, resulting in a recorded value of zero in the statistical analysis (Figure 1(E)).
Premature decrease in articular cartilage cell proliferation following intraperitoneal axitinib injections
According to the literature, the formation of the secondary ossification center (SOC) during early postnatal development influences growth plate chondrocytes and plays a key role in establishing the stem cell niche within the growth plate. 17 Inhibition of SOC maturation has been shown to delay the establishment of monoclonality in the growth plate.17,42,43 A negative correlation between SOC size and chondrocyte proliferation in the femoral articular cartilage of mice was observed (Figure 1(E)); however, the precise role of SOC in regulating articular chondrocyte proliferation during joint postnatal development remains unclear. To determine whether SOC formation affects cell proliferation in growing articular cartilage, SOC formation was delayed by administering axitinib (Figure 1(B)), a potent and selective inhibitor of VEGFRs. 44 Axitinib is an FDA-approved agent for the treatment of various malignancies by blocking angiogenesis.45,46 Axitinib successfully prevented vascular invasion in epiphyseal cartilage delaying the endochondral ossification.41,47 –49 In mice, continuous injection of axitinib within a narrow timeframe during postnatal development was shown to delay SOC maturation.17,42,43
Following axitinib injections from P10 to P13, Safranin O/Fast Green staining of joint tissue sections confirmed a delay in the SOC formation at P14 and P20 compared to control (Figure 1(D)). No changes in the number of EdU-positive (proliferating) cells in articular cartilage were observed in axitinib group at P14 (Figure 1(E)), however, at P20, the number of EdU-labeled cells in the axitinib group dropped significantly compared to the control (Figure 1(E)). In axitinib group, the number of EdU-labeled cells at P20 and P31 was comparable (Figure 1(E)). By postnatal day 31 in the completion of SOC development, there had been no longer any significant difference in the number of proliferating chondrocytes between axitinib group and control, both had a similarly low number of EdU-labeled cells (Figure 1(Dc-c″) and (E).
In summary, following the last axitinib injection at P13, no statistically significant effect on cell proliferation was observed at P14; however, a significant decrease in the number of EdU-positive cells was detected at P20. The effect of axitinib on chondrocyte proliferation after in vivo treatment can’t be considered direct, since, firstly, the half-life of axitinib in the body is 6 h, and secondly, the direct effects on chondrocytes in vitro and ex vivo showed additional negative effects. In experiments with 2D primary chondrocyte cultures and cultured explants of femoral epiphyseal cartilage, the direct effect of axitinib manifested as proliferation arrest and mitotic catastrophe (Supplemental Figure S2A–D), which is consistent with the literature.50 –52
AI-guided endogenous FLIM analysis of cells in the living tissue of articular cartilage revealed differences in FDPs in both the chondrocytes from different cartilage zones and the same cartilage zone of age groups P14 and P31
The conversion from proliferating juvenile chondrocytes to non-dividing adult chondrocytes should be accompanied not only by morphological changes in articular cartilage, but also by changes in the metabolic patterns of cells. To evaluate these changes occurred between the state of chondrocytes at P14 and P31, FLIM was performed with excitation and emission parameters optimized for mapping the FDPs of the NADH, an endogenous fluorescent marker sensitive to metabolic changes. Employing an AI-assisted approach for the automatic segmentation of FLIM images, the average fluorescence decay values (fluorescence lifetimes and amplitudes of the decay components) were quantified for individual cells, considering their geometric positioning relative to the cartilage surface (details in Methods) enabling us to determine with high accuracy the correlation between the FDPs and the depth of chondrocytes in cartilage. Using the distances of segmented cells from the surface, as well as visual features that distinguish cells from different zones of cartilage tissue (Figure 2(A) and (B)), the SF, TR, and DP zones of cartilage tissue were delineated, and changes in FDP were analyzed within each zone.

FLIM parameters alterations in articular cartilage cell zones in the living tissue of juvenile mice knee. (A) Fluorescence images of knee living sections at P14 and P31 merged with the color map of surface distance. (B) The plot of cell diameter alterations with distance from articular cartilage surface. (C and D) The cell distribution in the superficial, transitional, and deep zones of articular cartilage tissue at P14 (C) and P31 (D). (E) Fluorescence images showing the localization of markers for superficial zone cells (Cd73) and markers for hypertrophy (Mef2c) in articular cartilage in different age groups. (F and G) The representative images of the average fluorescence decay time τm on living femoral articular cartilage sections at P14 (F) and P31 (G) (scale bar, 50 µm). The dotted lines indicate the boundaries of the articular cartilage zones. (H and I) τ2 value dividing into groups based on the cell segmentation result value. (J and K) Amplitude ratio of short- and long-living components a1/a2 parameters distributing into groups of comparison based on the cell segmentation results. Data in (H–K) are presented as box plots showing the median and quartiles, and the 5-95 percentile (whiskers). (L) The distributions of τ2 value with distance from articular cartilage surface. (M) The distributions of the amplitude ratio of short- and long-living components a1/a2 parameters with distance from articular cartilage surface. Data were collected from 5-8 fields of view per mouse, and two joints per mouse; n = 4–5 different mice per group.
The approximate boundary between the SF and TR zones was set at 15 μm, 53 and the boundary between the TR and DP zones was defined at 60 μm from the articular cartilage surface. DP zone was settled within 60–150 µm and 60–110 µm in the P14 and P31 groups, respectively (Figure 2(C) and (D)). The relevancy of the established widths of cell zones in articular cartilage was confirmed on fixed joint sections by antibody staining against the markers of the progenitors (Cd73) inhabiting the SF 54 and hypertrophic chondrocytes (Mef2c) inhabiting the DP zones 55 (Figure 2(E)). Zone between Cd73+ and Mef2c+ cells is declared as TR.
Along with the gradient of cell morphology changes, FDPs demonstrated alterations with depth of tissue: according to the representative maps of the mean fluorescence lifetime (τm) for femoral cartilage sections in Figure 2(F) and (G), the fluorescence decay times τm in the DP zone of the cartilage became shorter in comparison to the τm measured in the SF zone in both groups. In fact, two FDPs are responsible for the τm behavior: the ratio of the amplitudes of the fast and long-living components a1/a2 and the fluorescence decay lifetime τ2. Changes in the protein binding or in the protein composition to which NADH binds affect the τ2 value. The value of a1/a2 is determined by the change in the ratio of the free and bound forms of NADH and is less affected by changes in the protein composition. 23 τ2 (Figure 2(H), (I), and (L)) and a1/a2 (Figure 2(J), (K), and (M)) parameters were analyzed separately, grouped according to the cell segmentation results (details in Methods).
The difference of τ2 was found to be statistically significant in cells belonging to different zones in both the P14 and P31 groups (Figure 2(H), (I), and (L)). In the P14 group, the median τ2 value gradually increased from 2.35 ns in the cells of SF zone to 2.40 ns in the cells of TR zone, and then further increased to ~2.50 ns in hypertrophic chondrocytes of DP zone (Figure 2(H), (I), and (L)). In the P31 group, the median τ2 values were higher in the cells of the SF zone (median τ2 = 2.81 ns) compared to the cells of another two zones (Figure 2(I)). At the same time, the statistically significant differences of τ2 values between TR and DP zones (median τ2 = 2.64 and 2.70 ns, respectively) were not found at P31 (Figure 2(I)). The different decay times τ2 apparently indicate the presence of different protein cofactors binding NADH in chondrocytes inhabiting different zones of articular cartilage in both age groups.
Similarly, the distributions of the a1/a2 values were compared and analyzed (Figure 2(J), (K), and (M)). In both P14 and P31 groups, the a1/a2 value demonstrated a tendency to increase with cartilage depth (Figure 2(K) and (M)). The median a1/a2 values measured at P14 in the SF, TR and DP zones were 2.28, 2.29 and 2.98 respectively. In the P31 group, the median a1/a2 ratio increased from 2.08 in SF to 2.44 in TR zones and then elevated to 2.6 in hypertrophic chondrocytes in DP zone. The pairwise comparison of median a1/a2 values showed significant differences between the P14 and P31 groups in two cellular zones, TR and DP (Figure 2(J)).
The variability of the median value of a1/a2 additionally characterized differences in the cell metabolic states in different zones of articular cartilage at P14 and P31 age. The observed increase in the a1/a2 value with depth (>50 µm from the surface for P14 group, and >10 µm for P31) (Figure 2(M)) reflects a shift of the cell state toward the predominance of the glycolytic pathway for ATP synthesis in deep layers of cartilage. At the same time, the area of proliferating cells in P14 is much wider in group P14 than in P31 (Figure 1(Cb″) and (Cd′)). As well, P14 had a larger area of Cd73-positive cells (wider SF zone) than P31 (Figure 2(E)). This could explain the earlier and steeper increase in the a1/a2 ratio with increasing distance from the cartilage surface in the P31 group compared to P14 (Figure 2(M)).
To ensure that the observed differences in measured FDPs were not attributable to oxidative stress resulting from disruption of the synovial capsule integrity during articular cartilage extraction, ROS generation was assessed in living cartilage tissues following isolation of knee joint epiphyses at the P14 and P31 developmental stages (Supplemental Figure S2E). The presence of superoxide radicals (O2•⁻), the predominant ROS in chondrocytes, 56 was evaluated using nitroblue tetrazolium (NBT) staining 57 of isolated femoral epiphyses. Both articular cartilages from P14 and P31 mice exhibited no detectable staining (Supplemental Figure S2E), in contrast to positive control samples pre-treated with menadione, a compound known to rapidly induce oxidative stress at multiple cellular sites. 58
Assessment of the distribution of metabolic enzymes across different zones of juvenile and adult articular cartilage via immunofluorescence staining
In addition to measurements of FDPs, the distribution of glycolytic factors in articular cartilage during postnatal joint growth, including lactate dehydrogenase A (Ldha), the mitochondrial enzyme citrate synthase (Cs), and pyruvate dehydrogenase (Pdh) phosphorylated (pPdh), was assessed (Figures 3(A), (B) and 4(A)–(C)). They are key enzymes regulating the redistribution of glucose flow between glycolysis and OXPHOS. Quantitative analysis of fluorescence intensity of articular cartilage sections at P5, P14, P20, P31, and P65 demonstrated that Ldha fluorescence increased dramatically with the depth of chondrocyte position in the cartilage. This fluorescence distribution, observed consistently from the SF zone to the DP zone across all age groups analyzed (Figure 3(Ba) and (Bc)), aligns with elevated pPdh levels detected in the DP zone (Figure 4(B)). Phosphorylated Pdh fluorescence remained relatively unchanged in the SF and TR zones across all age groups (Figure 4(A)–(C)). The DP zone had the highest pPdh fluorescence level compared to the SF and TR zones in all age groups (Figure 4(B)), while a gradual decrease in pPdh fluorescence in the DP zone with age was observed (Figure 4(C)).

Immunofluorescence staining with antibodies to the metabolic enzymes Ldha and Cs and analysis of fluorescence intensity. (A) The observation time are postnatal days 5, 14, 20, 31 and 65. Representative images of study groups are displayed, magnification ×200. Confocal LSM images of 30 μm sections are represented as Z-axis intensity projections. The yellow dotted lines indicate the boundaries of the articular cartilage zones. The negative control is presented in the bottom row of panel A. (B) Fluorescence intensity was calculated as the mean gray value of the individual cell in age groups (Ba, Bb) and in cell zones of articular cartilage tissues (Bc, Bd). Data are presented as box plots showing the median and quartiles, and the 5-95 percentile (whiskers). Data were collected from one field of view per section, three tissue sections per mouse, one joint per mouse, and three mice (n = 3) per group. To assess fluorescence intensity of Ldha and Cs, the mean number of cells measured per section (k̄) was as follows: k̄SF = 22, k̄TR = 30, k̄DP = 27.

(A–C) Immunofluorescence staining with antibodies targeting the phosphorylated form of Pdh (pPdh), followed by quantitative analysis of fluorescence intensity. (A) Representative images of the study groups are displayed, magnification ×200 (scale bar, 100 µm). The observation time are postnatal days 5, 14, 20, 31, and 65. Confocal LSM images of 30 μm sections are represented as Z-axis intensity projections. The yellow dotted lines indicate the boundaries of the articular cartilage zones. (B and C) Fluorescence intensity was calculated as the mean gray value of the individual cell in age groups (B) and in cell zones of articular cartilage tissues (C). Data are presented as box plots showing the median and quartiles, and the 5-95 percentile (whiskers). Data were collected from one field of view per section, three tissue sections per mouse, one joint per mouse, and three different mice (n = 3) per group. To assess fluorescence intensity of pPdh, the mean number of cells measured per section (k̄) was as follows: k̄SF = 22, k̄TR = 29, k̄DP = 27. (D) Safranin O/Fast Green staining of the knee joint sections collected on day 15 after osteochondral defect induction performed at P14 and P31. Original magnification: ×100.
In contrast, the differences in Cs fluorescence across the zones for each age group were less pronounced (Figure 3(Bb) and (Bd)). Despite small fluctuations in median values of Cs fluorescence, the levels of Cs in DP zone were relatively stable among age groups (Figure 3(Bd)). The Cs fluorescence level in cells of the SF zone of articular cartilage was significantly decreased at P65 compared with P5 and P20 (Figure 3(Bd)). Although the P14 and P31 groups showed a similar trend toward a difference from P65, these differences did not reach statistical significance (Figure 3(Bd)). Note that chondroprogenitor cells in the SF zone of adult articular cartilage at P65, which exhibited reduced Ldha fluorescence levels, also showed a trend toward the lowest median fluorescence levels of Cs and pPdh (Figures 3(B), 4(B) and (C)). Low fluorescence levels of all three enzymes in the SF zone at P65, compared to SF cells from younger mice, likely indicate reduced metabolism in the chondroprogenitor cell population in adult articular cartilage.
Juvenile mouse chondrocytes exhibiting high proliferative activity are unable to effectively repair osteochondral defects in vivo
According to our data, chondrocyte division showed a gradual and uniform decrease between P5 and P31 (Figure 1(C) and (E)). It was investigated whether cells capable of proliferation provide a more effective regenerative response to acute injury than adult cartilage chondrocytes. An osteochondral defect was formed in the femoropatellar groove of the mouse knee joint in two age groups: at P14 and P31. Two weeks and 1 month were chosen as the time points when chondrocytes exhibited high proliferative activity and when it faded simultaneously with the completion of femoral SOC formation respectively. Joint tissues for analysis were taken in 15 days after surgery. According to Safranin-O/Fast Green staining, restoration of the native structure of articular cartilage was not observed in both P14 and P31 groups on day 15 after surgery (Figure 4(D)). Despite the osteochondral defect formed at P14, the development and maturation of articular cartilage surrounded the defect was not disturbed (Figure 4(D)).
Discussion
Articular cartilage chondrocytes gradually reduce their proliferative activity during postnatal joint growth and become quiescent at maturity or following axitinib injections
Joint cavity and articular cartilage surface formation occurs during embryonic joint development. 59 In mice, joint maturation continues postnatally, culminating in mature articular cartilage and SOC formation.1,43,60 During the juvenile period, articular cartilage tissue turnover occurs, and chondrocytes of fetal origin are replaced by the progeny of cells located in the superficial zone of articular cartilage.5,16 After postnatal day 30, when SOC formation completes, articular cartilage turnover nearly ceases, 5 and dividing cells decline progressively with age (highest EdU-labeling at P5, lowest at P31 (Figure 1(E)). Chondrocytes divide actively in the first postnatal weeks but enter quiescence by ~P31, coinciding with SOC completion. Our study shows that, contrary to abrupt cessation, chondrocyte proliferation declines steadily from birth to adulthood (P65). At P65, the cessation of cell division may explain the absence of EdU-positive cells, or the cells may have divided so rarely that a 3-day EdU pulse was insufficient to detect this effect.
Previous studies 61 have demonstrated that the growth factor VEGF and its receptors (VEGFRs) are actively expressed in juvenile chondrocytes during articular cartilage growth. In growing cartilage, VEGF and VEGFRs are produced by chondrocytes within the SF and TR zones, which correspond to regions of active cell proliferation. 61 With maturation, VEGF expression is downregulated in chondrocytes, paralleling the decline in their proliferative activity. 61
Axitinib, a potent VEGFR inhibitor with established antiproliferative effects in cancer cells, inhibits vascular invasion into postnatal developing joints by disrupting endochondral ossification. That remodels the epiphyseal plate cartilage without affecting its growth, while also impairing SOC development and joint maturation.17,42 Despite axitinib’s established clinical utility and existing data regarding its effects in alternative contexts, its influence on juvenile articular chondrocyte proliferation had remained unexamined prior to the present investigation.
Axitinib is characterized by a relatively short half-life (6 h) and rapid systemic clearance, 62 therefore, within the joint, it is likely to primarily target vascularized tissues. The present study shows that axitinib administration at P10-P13 has no effect on chondrocyte proliferation in articular cartilage at P14—1 day after the last injection—no effect on chondrocyte proliferation in articular cartilage was observed; the inhibitory effect became apparent only later, at P20. In contrast, during in vitro and ex vivo experiments, juvenile chondrocyte division ceased the day after axitinib application.
It is logical to conclude that the delayed effect of axitinib on juvenile chondrocyte proliferation in vivo is indirect. The direct inhibitory action of axitinib primarily targets vascular tissues such as SOC and the synovial membrane. Alterations in these vascularized tissues subsequently influence the development of articular cartilage. Another plausible explanation is that the reduction in chondrocyte proliferation resulted from delayed joint development induced by axitinib administration during the early postnatal period. This, in turn, subsequently led to a desynchronization between the intrinsic processes of joint development and external systemic signaling during later developmental stages. Axitinib’s direct inhibition of vascularized structures like the SOC and synovial membrane likely disrupts signals or environmental conditions necessary for normal articular cartilage development, which in turn reduces chondrocyte proliferation.
Postnatal immature articular cartilage fails to restore full-thickness injury
The limited repair capacity of mature articular cartilage is well established.63,64 This failure stems primarily from reduced chondrocyte proliferation, minimal collagens turnover,65,66 and the tissue’s avascular nature. Prior studies, 67 show that full-thickness chondral defects in 2–3-month-old mice fail to restore native structure, and cell lineage tracing techniques5,16,67 revealed a dramatic slowdown in tissue turnover in articular cartilage after P30. Although 2-week-old mouse chondrocytes proliferate more actively in vivo than those from 1-month-olds, our results demonstrate that these juvenile cells cannot restore original tissue architecture following full-thickness injury. In contrast, previous work reported spontaneous repair of articular cartilage injuries in postnatal immature rats and rabbits.68 –70 However, those studies used partial-thickness linear cartilage injury (PTI) models—shallow, narrow lesions not reaching the tidemark of articular cartilage—explaining the discrepancy with our full-thickness defect regeneration data.
A full-thickness articular cartilage injury (FTI) model was used to study fetal articular cartilage regeneration in sheep. 71 The study confirmed fetal regenerative versus adult scarring cartilage repair and illustrated that no proliferating cells (demonstrated by the expression of the proliferation antigen Ki67) were found in the articular cartilage of adult animals, and in fetal healthy cartilage, Ki67-positive cells were detected throughout the whole cartilage. 71 Fetal scarless regeneration represents a paradigm for ideal tissue repair. The regenerative potential of grafts from interzone (IZN) and adjacent anlagen (ANL) (tissues preceding the formation of the joint in fetus) was analyzed in FTI rat model. 72 Defects treated with equine fetal IZN, ANL, or chondrocyte pellets developed hyaline cartilage with a zonal chondrocyte arrangement and extracellular matrix content like that of native cartilage, however, IZN and ANL implants showed overgrowth.
Nevertheless, according to our study, postnatal immature articular cartilage of mice was not restored 15 days after femoral osteochondral defect. This is probably due to the specificity of cartilage tissue growth driven by apposition of daughter cells from surface to deep zone directed perpendicular to the surface of the joint epiphysis (appositional growth). 67
The differences in metabolic patterns between juvenile and adult articular chondrocytes
Due to avascular nature of hyaline cartilage, its cells obtain glucose and oxygen from the synovial fluid and have the ability to adjust to low-oxygen conditions. 7 In our study, by assessing FLIM parameter a1/a2 ratio between SF, TR and DP zones in articular cartilage, a significant metabolic shift was observed. The widespread interpretation of the increase of the a1/a2 parameter is the shift of cell metabolism from OXPHOS state to glycolysis.73 –75 According to FLIM data, a1/a2 ratio showed a rise in value in cells with increasing distance from the cartilage surface. On the one hand, inherently, this may be partly a consequence of the fact that cells from different zones are exposed to unequal conditions of oxygen input. The cellular consumption of oxygen depletes their concentration inside the articular cartilage tissue, creating the gradient of oxygen concentrations from the SF zone of cartilage toward the subchondral bone, 76 and the oxygen deficiency in cartilage depth causes cells to prefer glycolysis for ATP production. This assumption is supported by the fact that cells in the SF zone, which are in close contact with the synovial fluid, showed no significant difference in median a1/a2 values between the P14 and P31 groups. The synovial fluid, the cartilage source of nutrients and oxygen bathing the superficial cartilage, is known to be a plasma filtrate with glucose and oxygen concentrations not exceeding those of blood, 21 and the synovial fluid concentration is assumed to be more or less constant. In groups P14 and P31, the median values of a1/a2 of non-proliferating cells in the DP zone lying on the subchondral bone were different, which could probably be a consequence of the differences in the availability of oxygen and glucose conditioned by thickness of the articular cartilage tissue, which becomes thinner with age.
On the other hand, an increase in the a1/a2 ratio with cartilage depth in both age groups indicated that the changes in chondrocytes from chondroprogenitors of the SF zone to differentiating chondrocytes associated with the TR zone and hypertrophic chondrocytes in the DP zone was accompanied by a shift of metabolic patterns. During the maturation, the chondrocyte morphology change and size increase are known to require an influx of building elements, providing by glycolysis. 18 Circumstantial evidence can be found in the published articles. Oxygen availability appears to be one of the external conditions influencing the sequential process of chondrocyte differentiation in vivo. The expression of lubricin, a marker of chondroprogenitor cells mainly produced by chondrocytes in SF zone,5,77 shows significant decrease under hypoxic condition. 78 In the DP zone, healthy articular cartilage is characterized by a strong intracellular level of the oxygen-sensitive subunit alpha of the transcription factor hypoxia-inducible factor 1 (HIF-1α),79,80 the chief mediator of hypoxic response in mammalian tissues. 81 HIF-1α in turn reduces cell proliferation 82 and increases the transcription of glucose transporters, many glycolytic enzymes, and Ldha.83,84 In the cartilaginous growth plate, HIF-1α is required for survival of hypoxic chondrocytes, and it is expressed in hypertrophic chondrocytes.18,41
Comparison of another FLIM parameter τ2 in age groups P14 and P31 revealed a significant difference in metabolic states in articular cartilage chondrocytes. In the P31 group, compared with the P14 group, τ2 values were significantly higher regardless of the cartilage cell zone, even in SF zone chondrocytes. In FLIM method, NADH fluorescence lifetime of the long-lived component τ2 provides a measure of the NADH microenvironment, mostly protein binding, accordingly, τ2 was influenced by the type of cofactors binding to the NADH molecules.85 –87 The NADH cannot be separated spectrally from coenzyme nicotinamide adenine dinucleotide phosphate (NADPH) autofluorescence. 88 The measured FDPs are always a mixture of both NADPH and NADH autofluorescence. 88 The concentrations of the NAD(H) redox pair are about 10 times the levels of NADP(H), nevertheless NADPH contributes to the measured autofluorescence parameters.23,89 The NADH molecule is involved in OXPHOS and glycolysis, and NADPH is a product of the pentose phosphate pathway (PPP) and is involved in the ROS defense. 90 In most healthy mammalian cells in unstressed conditions, the PPP exhibits much lower flux than glycolysis, however, the PPP can be transiently activated to meet urgent NADPH demands. 91
The reducing power that is required for the synthesis of lipids, deoxyribonucleotides and proline is provided by NADPH, and the largest source of NADPH in the cytosol is the PPP. 90 NADPH also contributes to the recycling of oxidized glutathione to reduced form, which scavenges ROS, 92 the toxic by-products of normal cellular metabolism.
ROS are the one of the main factors of oxidative stress impairing the metabolism of chondrocytes. When the balance between ROS generation and elimination is destroyed, oxidative stress occurs and ROS continually attack lipids, proteins, and DNA, resulting in severe and irreversible oxidative damage. 93 Oxidative stress has long been considered a significant factor in the induction of chondrocyte senescence and apoptosis in diseases such as OA. Chondrocytes, similar to other mammalian cells, have an antioxidant system to eliminate ROS and maintain redox balance. 93
Physiological levels of ROS support intracellular signaling in chondrocytes, the balance between ROS production and elimination ensures that redox-sensitive signaling proteins function correctly. 93 Being the primary source of ROS in cells, mitochondria appeared to be a major target of ROS production, however, ROS are also generated in the endoplasmic reticulum (ER) when disulfide bridges are formed during oxidative protein folding. 94 Previous studies have also identified LDHA-mediated ROS production in vitro and in vivo independent of its canonical lactate-producing role.95 –97 LDHA binding to NADH in a cell-free system can greatly increase the rate of oxidation and free radical generation.95,98,99 It is likely that the interaction of NADH with a special domain of LDHA amplifies the donation of electrons from NADH to oxygen-containing compounds through its catalytic activity, as well as by thermodynamic stabilization of free radical intermediates.100,101
Generation of ROS by NADPH oxidases (Nox) is thought to be important for signal transduction in chondrocytes. 93 Among Nox family members, Nox1, Nox2, and Nox4 are required for ROS generation in chondrocytes during chondrogenesis. 102 The Nox1 and Nox2 content in chondrocytes is increased as differentiation progressed, whereas that of the Nox4, which is initially high, was gradually decreased during differentiation.102,103 It has also been shown that the initiation of chondrocyte hypertrophy is modulated by intracellular Nox-dependent ROS generation. 104
NAD(P)H is involved in ROS generation and elimination pathways modulating chondrocyte differentiation and hypertrophy. 93 Alterations in cellular growth rate could influence NADPH levels via regulation of PPP producing critical precursors to synthesize various biomolecules. 105 Changes in biological processes associated with NADPH could be reflected by the prolonged long-lifetime component τ2 value.23,89
The measured differences in τ2 value found between 2-week-old and 1-month-old cartilages indicate a redistribution of glucose flux between main metabolic pathways involving NAD(P)H molecules such as OXPHOS, glycolysis, and PPP. Different metabolic profiles of chondrocytes were due to different functional priorities of cartilage at P14 and P31 ages. At the age of 2 weeks (P14), cartilage development continued, and chondrocytes required to divide and produce the extracellular matrix. 1 The primary functional task of adult articular cartilage (over 30 days old) was to perform the physical and mechanical function of absorbing joint movement, which was accompanied by a stop in chondrocyte proliferation. 1 It is important to remember that this is also accompanied by a process of chondrocyte differentiation, which is active during adolescence and is superimposed on other metabolic changes.
The metabolic enzyme production changes with articular chondrocyte maturation
The hallmarks of hyaline cartilage are its hierarchical structure, avascular nature and lower level of oxygen content and consumption compared to the other tissue types. 76 According to the literature, anaerobic glycolysis is the preferred metabolic pathway for chondrocytes in adult healthy cartilage to obtain energy and it plays an important role in maintaining normal function and preventing degeneration of cartilage. 106 For adult cartilage chondrocyte culture, it was shown that mitochondrial respiration in adult cartilage is not limited by oxygen availability, because even at normoxic oxygen concentrations chondrocytes continue to primarily use glycolysis for ATP production. 107 In chondrocyte cultures, replacing glucose with galactose as the sole sugar source leads to a reduction in Ldha activity and lactate production, and an increase in basal oxygen consumption without increasing mitochondrial content. 107 During postnatal joint development, the articular cartilage tissue undergoes tremendous structural changes over postnatal life.1,67 At the same time, articular cartilage cells, silent in adulthood, actively proliferate in juvenile age, which requires additional energy expenditure. A special relationship exists between glucose metabolism and rapid cell proliferation 13 : while maintaining homeostasis, actively proliferating cells have increased energy and intermediates requirements for growth and division.13,14,18 It is widely known that some cell types, such as embryonic and cancer cells, tend to metabolize glucose anaerobically rather than aerobically, even when oxygen is available (so called the “Warburg effect” or “aerobic glycolysis”).20,108,109
Given that the balance between glycolysis and oxidative metabolism is regulated by metabolic enzymes such as Ldha, Cs, as well as Pdh, 110 the expression of Ldha and Cs as well as an inactivated form of Pdh, pPdh, was mapped in articular cartilage tissue sections by immunofluorescent staining. In OXPHOS, Cs is generally assumed to be the rate-limiting enzyme of the tricarboxylic acid (TCA) cycle, which regulates energy generation in mitochondrial respiration. 111 Pdh converts pyruvate obtained from glucose into acetyl coenzyme A. 110 In the deficiency of oxygen, pyruvate dehydrogenase kinase inactivates Pdh by phosphorylation, and pyruvate is processed by Ldha into lactate, the final product of anaerobic glycolysis.
In articular cartilage, an increase in the presence of glycolysis enzyme Ldha with depth was observed in all age groups, which is consistent with the FLIM data obtained. An increase of Ldha and pPdh, proteins facilitating anaerobic metabolism, was detected in non-proliferating hypertrophic chondrocytes in DP zone compared to cells from other zones. Similar findings were previously done using the chondrocytes of growth plate, another joint hyaline cartilage. High levels of Ldha in hypertrophic chondrocytes in growth plate cartilage were found as a result of the expression of transcription factor HIF1α initiated by hypoxia. 112 Following treatment of the growth plate with reagent disrupting oxidative phosphorylation, a significant decrease in ATP generation was noted only in the proliferating zone, but not in hypertrophic chondrocytes. 113 The elevated reliance of hypertrophic cells mostly on anaerobic metabolism is quite understandable. During the first month after birth, not only dividing processes may require energy expenditure. The several times increase in volume in the early hypertrophic chondrocytes accompanied by production of cellular dry mass 114 requires a significant input of building blocks (e.g. amino acids and lipids) produced by anaerobic glycolysis.12 –14
During the first month of postnatal joint development, Ldha levels showed a trend toward a slight increase with age, while Cs distribution did not exhibit significant variation across age groups in the TR and DP zones of articular cartilage. However, in the SF zone, noticeable changes in Cs level were observed. The Cs level in the SF zone showed a steady decrease with age, although this trend was not statistically significant. However, at P65, the Cs level reached a significantly lower value compared to P5. Apparently, with a decrease in proliferative activity, the main function of chondroprogenitors, the energy requirements of cells reduce, and the progenitor cells of SF zone put down the contribution of OXPHOS to their metabolism.
At the P65 stage, only single, very rare proliferating cells were present in the articular cartilage tissue. At that time, the enzymes Ldha, Cs, and pPdh exhibited their lowest levels in the SF zone of the articular cartilage. Chondroprogenitors in adult articular cartilage are thought to decrease not just proliferation, but also metabolic activity. Energy-intensive processes, including differentiation, proliferation, and synthesis of extracellular matrix, in adult articular cartilage are suspended in the SF zone cells, therefore, the common metabolism level of slow-dividing chondroprogenitor cells can be extremely low. However, this is not applicable to the chondrocyte population of the TR and DP zones, which stop to proliferate and differentiate, but retain their function of the extracellular matrix turnover and maintaining homeostasis in adult articular cartilage and persist common metabolism on the higher level compared to the SF zone progenitors. Not only articular cartilage cells of different ages have significant differences in metabolic states, but also chondrocytes of the same tissue, which inhabit different layers of cartilage. Chondrocytes are likely to manage glucose fluxes to meet the energy needs of processes such as proliferation and differentiation in situation where oxygen and glucose availability are restricted.
Conclusion
Chondrocytes of articular cartilage can be either actively dividing in juvenile period or silent in adult, but they remain functional chondrocytes in both cases. Using the FLIM method, it was shown that articular cartilage cells alter their metabolic pattern during the postnatal period, depending on cell maturation stage and mouse age. Functional changes of chondrocytes are probably a reason for metabolic reprogramming and a shift in their energy supply balance. Cell division and production of extracellular matrix are necessary for the development of juvenile cartilage. The primary functional task of adult articular cartilage (over P30) is to perform the mechanical function of supporting joint movement. 1 The present study showed that the maturation process of chondrocytes is accompanied by metabolic changes, and the contribution of glycolysis to cellular metabolism increases with the increasing depth of cartilage layers. According to their needs, the juvenile proliferating chondrocytes orchestrate the OXPHOS, glycolysis, and PPP activity. Chondrocytes are challenged by coordinating glucose fluxes to support the energy requirements for proliferation and hypertrophy, as well as effective energy-demanding production of glucose-derived matrix glycosaminoglycans. The metabolism in chondrocytes adapts to different stages of chondrocyte maturation and changes in postnatal articular cartilage development.
To determine the distribution of oxygen and glucose in tissue, it’s important to take into account more than just the thickness of the cartilage tissue and the density of consuming chondrocytes. The cell proliferative state and cell differentiation stage have an influence on metabolic processes. Today, scientists do not delimit data obtained from juvenile and adult chondrocytes in their studies. However, metabolic differences between early postnatal and adult chondrocytes may influence the results of the studies conducted, therefore, caution should be exercised when generalizing data obtained from juvenile chondrocytes to the capabilities of adult chondrocytes and vice versa.
It should be noted that murine joints do not fully recapitulate human joints in terms of anatomy, including continuously open growth plate, as well as biochemical characteristics and biomechanics. Therefore, the results gained on mice allow limited clinical translatability. 15 Nevertheless, conducting more research using animal models to determine the biological mechanism responsible for managing the switch from active proliferative to adult cartilage silent state and understanding the role of glucose metabolism in controlling these processes in chondrocytes is necessary for the joint regenerative medicine. It is still unclear whether changes in the metabolic state of chondrocytes are a consequence of cellular changes caused by their maturation and whether it is possible to control the proliferation and/or differentiation of chondrocytes through the regulation of their metabolic processes.
A deeper understanding of the processes that control the ability of chondrocytes to proliferate and/or differentiate may address limitations caused by the insufficient number of autologous chondrocytes derived from patients, and the effectiveness of developing cartilage bioengineering and cell therapy for joint restoration could be significantly improved.
Supplemental Material
sj-png-1-tej-10.1177_20417314261432891 – Supplemental material for Chondrocyte metabolic transition from proliferation to quiescence revealed by FLIM in postnatal mouse knee joints
Supplemental material, sj-png-1-tej-10.1177_20417314261432891 for Chondrocyte metabolic transition from proliferation to quiescence revealed by FLIM in postnatal mouse knee joints by Nadezda Ignatyeva, Boris Yakimov, Anastasiia D Kurenkova, Irina A. Romanova, Pavel D. Kibirskiy, Nikita Gavrilov, Anastasiya Patsyurkevich, Irina D. Shcherbakova, Artem M. Mozherov, Aleksandra V. Kashina, Evgeny A. Shirshin, Peter S. Timashev and Ekaterina V. Medvedeva in Journal of Tissue Engineering
Supplemental Material
sj-png-2-tej-10.1177_20417314261432891 – Supplemental material for Chondrocyte metabolic transition from proliferation to quiescence revealed by FLIM in postnatal mouse knee joints
Supplemental material, sj-png-2-tej-10.1177_20417314261432891 for Chondrocyte metabolic transition from proliferation to quiescence revealed by FLIM in postnatal mouse knee joints by Nadezda Ignatyeva, Boris Yakimov, Anastasiia D Kurenkova, Irina A. Romanova, Pavel D. Kibirskiy, Nikita Gavrilov, Anastasiya Patsyurkevich, Irina D. Shcherbakova, Artem M. Mozherov, Aleksandra V. Kashina, Evgeny A. Shirshin, Peter S. Timashev and Ekaterina V. Medvedeva in Journal of Tissue Engineering
Footnotes
Acknowledgements
The work was performed using the equipment of the Shared Use Center “Center for Laser Technologies in Medicine” with the support of the Ministry of Science and Higher Education of the Russian Federation (Agreement No. 075-15-2025-669 dated August 5, 2025). FLIM analysis was combined with the AI-assisted original approach for the automatic segmentation of FLIM images based on the use of the zero-shot Segment Anything Model (SAM) with pre-trained weights. The authors declare that all results and images obtained are original and that the study is not AI-generated.
List of abbreviations
AI, artificial intelligence; ANL, anlagen adjacent to the interzone; ATP, adenosine 5′-triphosphate; EdU, 5–ethynyl–2′–deoxyuridine; Cs, citrate synthase; DP, deep; FDPs, fluorescence decay parameters; FLIM, fluorescence-lifetime imaging microscopy; FTI, full-thickness articular cartilage injury; IZN, interzone; Ldha, lactate dehydrogenase A; LMM, linear mixed-effects model; Mef2c, myocyte-specific enhancer factor 2C; NADH, nicotinamide adenine dinucleotide reduced form; NADPH, nicotinamide adenine dinucleotide phosphate; NBT, nitroblue tetrazolium; OXPHOS, oxidative phosphorylation; Pdh, pyruvate dehydrogenase α1; pPdh, pyruvate dehydrogenase α1 phosphorylated; PPP, pentose phosphate pathway; PTI, partial-thickness linear cartilage injury; ROS, reactive oxygen species; SAM, segment anything model; SF, superficial; SOC, secondary ossification center; Sox9, SRY-box transcription factor 9; TR, transitional; VEGF, vascular endothelial growth factor; VEGFRs, vascular endothelial growth factor receptors.
Ethical considerations
Experiments on animals were conducted in the vivarium at the Sechenov University in accordance with the European Convention (Strasbourg, 1986) and the World Medical Association Declaration of Helsinki on the human treatment of animals (2000). The animal experiments were performed in the frame of the project “Testing of cell therapies and tissue-engineered constructs for compensating joint injuries of various etiologies” approved by the Local Ethics Committee of the I.M. Sechenov First Moscow State Medical University (Sechenov University) (Protocols No. 22-01-e of 29 December 2022, 22-03-e, and 22-04-e of 28 February 2023, Moscow, Russia).
Author contributions
Conceptualization, EVM, AVK, and EAS; Formal analysis, BY, NI, AMM, PDK, and EAS; Funding acquisition, PST and EVM; Investigation, IAR, ADK, PDK, IDS, AMM, NI, NG, AP, and EVM; Methodology, EVM, NI, IAR, BY, ADK, PDK, and EAS; Project administration, PST; Supervision, EVM, AVK, EAS, and PST; Writing—original draft, EVM, BY and EAS; Writing—review & editing, EVM, NI, BY, and EAS. All authors have read and agreed to the published version of the manuscript.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The in vitro studies and experiment on osteochondral explants were conducted with the support of the Russian Science Foundation, project No. 25-74-00117,
. FLIM measurements and immunostainings of cartilage tissues were funded by the Russian Science Foundation, project No. 21-75-10082. The funders had no role in the design, analysis and reporting of the study.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
Representative selections of all datasets are included in the paper and Supplemental Materials. The full FLIM datasets are not publicly available due to their file sizes, but are available from the corresponding author upon reasonable request.*
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
