Amelogenin Null Mice Develop Osteoarthritis, While Its Application Mitigates Disease Phenotypes in a Rat Model

ABSTRACT

Previous studies have demonstrated that recombinant human amelogenin protein (rHAM⁺) promotes healing of injured articular cartilage, subchondral bone, and skeletal ligaments. Therefore, we speculated that amelogenin may play a role in osteoarthritis (OA) development. Aged amelogenin‐null and wild‐type mice underwent micro‐computed tomography (micro‐CT) and histological analyses to assess OA‐related changes. Additionally, OA was induced in rat knees via destabilization of the medial meniscus, followed by treatment with 0.5 mg/mL rHAM⁺ dissolved in propylene glycol alginate (PGA) or PGA alone. Magnetic resonance imaging (MRI) and histological analyses were performed. Twenty‐three‐month‐old amelogenin‐null mice exhibited severe OA features, including cartilage loss, joint space narrowing, and osteophyte formation, whereas wild‐type mice showed only mild, age‐related changes. OA pathology was evident in 12‐month‐old amelogenin‐null mice, by increased matrix metalloproteinase‐13 (MMP‐13) and decreased type II collagen expression. In osteoarthritic rats, MRI analyses demonstrated that treatment with rHAM⁺ delayed disease progression and improved OA phenotypes. Twenty‐four weeks posttreatment, the levels of type II collagen increased, while MMP‐13 and type X collagen decreased. MMP‐13 reduction was detected as early as 2 weeks posttreatment, contributing to cartilage preservation. Furthermore, similar to the known effect of rHAM⁺ in acute injuries, recruitment of CD105‐positive mesenchymal stem cells to the cartilage was detected 5 days posttreatment. Lack of amelogenin led to the development of osteoarthritic phenotypes, whereas in the induced osteoarthritis model, a single application of amelogenin inhibited joint deterioration and partially healed osteoarthritic damage compared with the control. These findings highlight the potential of amelogenin as a disease‐modifying agent for OA.

Aged amelogenin‐null mice developed severe osteoarthritic features. In a rat model of induced osteoarthritis, a single application of a recombinant human amelogenin inhibited joint deterioration and induced partial healing, while increasing type II collagen and decreasing MMP‐13 and type X collagen, 24 weeks posttreatment.

1. Background

Amelogenin constitutes approximately 90% of the extracellular enamel matrix proteins. It plays a pivotal role in biomineralization and structural assembly during enamel development, guiding the oriented and elongated growth of enamel mineral crystals. Enamel mineralization is facilitated by the sequential degradation of amelogenin proteins and their replacement with mineral ions [1]. In humans, mutations in the amelogenin gene, AMELX (NCBI nucleotide: NM_182680.1), cause the rare X‐linked form of Amelogenesis Imperfecta (AI) (NCBI OMIM: #301200), which most commonly affects enamel. Amelogenin knockout mice exhibit defective enamel formation similar to AI, with reduced enamel thickness due to disorganized hypoplastic enamel. Additionally, amelogenin‐null mice display defects in cementum, the thin mineralized layer covering the tooth root, along with decreased body weight [2, 3, 4, 5, 6]. Mice lacking only the AmelX phosphorylation site (Ser‐16) also fail to form enamel rods [7]. However, no reports to date have described non‐dental phenotypes in AI patients or amelogenin knockout mice.

During tooth formation, amelogenin is also expressed in additional tissues, including ectomesenchymal‐derived odontoblasts and periodontal ligaments [8, 9]; mesenchymal non‐dental tissues such as long bones and cartilage cells; and non‐mineralized tissues, including the brain, eyes, ganglia, and peripheral nerve trunks [10, 11]. Based on these findings, we hypothesized that amelogenin‐null mice might exhibit skeletal and joint abnormalities.

A single application of the recombinant human amelogenin protein (rHAM⁺), produced in our laboratory [12], induced healing of injured or diseased dental and musculoskeletal tissues. The regenerated tissues include periodontal tissues following experimentally induced periodontitis in a dog model [13]; acutely injured rat knee medial collateral and ankle calcaneofibular ligaments [14, 15]; and acutely injured rat hyaline articular cartilage and subchondral bone of the femoral trochlea [16]. The application of rHAM⁺ to the injured or diseased site induced recruitment of cells expressing mesenchymal stem cell (MSC) markers [13, 14, 16].

Osteoarthritis (OA), the most common degenerative joint disease, develops through a series of pathological changes. These include progressive loss and destruction of articular cartilage, thickening of the subchondral bone, osteophyte formation, synovial inflammation, degeneration of knee ligaments and menisci, periarticular muscle structural changes, and joint capsule hypertrophy [17]. OA causes pain, limited mobility, and joint deformities, significantly impairing quality of life and increasing the economic burden on aging populations [18]. Current OA treatments are aimed mainly at relieving inflammation, pain, and stiffness. These treatments include physical activity, topical creams, non‐steroidal anti‐inflammatory drugs (NSAIDs), steroidal drugs, and complementary medicines. However, in patients affected by severe OA, joint replacement or fusion is the only established surgical technique [17]. To date, there are no approved interventions existing to restore cartilage or curtail disease progression; therefore, OA presents a major scientific and clinical challenge.

Given amelogenin's ability to promote the healing of injured joint tissues, we speculated that it might influence OA development. Here, we elucidated the effects of amelogenin depletion on the skeletal joints in amelogenin‐null mice and evaluated the therapeutic potential of rHAM⁺ in a rat model of induced OA.

2. Methods

2.1. Animals

The experimental protocols were approved by the Authority of Biological and Biomedical Models (ABBM), The Hebrew University, Jerusalem, Israel (HUJI). The animals were housed under standard conditions of controlled temperature (22°C ± 2°C), 12‐h light/dark cycles, humidity (30%–70%), proper ventilation, and ad libitum access to food and water. Exclusion criteria were according to ethical approval and included clinical and/or behavioral changes. An ARRIVE checklist is provided in the Supporting Information.

2.2. Study Design

A schematic diagram of the study design is presented in Figure 1.

FIGURE 1.

Schematic diagram of the study design.

2.3. Amelogenin‐Null Mice

Four female mice heterozygous for the AmelX amelogenin‐null gene (B6;129‐Amelxtm1Kul/Unc) with a mean weight of 20 g and four wild‐type C57BL/6 male mice (C57BL/6JRccHsd) with a mean weight of 25 g were purchased from the Mutant Mouse Resource and Research Centers (MMRRC) and mated. F1 offspring were genotyped for the amelogenin‐null AmelX allele, and only F1 female carriers were further mated to obtain F2 mice homozygous for either the wild‐type (WT) or amelogenin‐null allele. Genotyping verification for wild‐type and knockout (KO) mice was conducted at the F3–F5 generations and subsequently every 6 months.

DNA for genotyping was extracted from an ear punch by incubating the sample at 94°C for 10 min in 10 mM NaOH, followed by cooling to room temperature and neutralization with 1:10 (v/v) 0.1 mM Tris‐EDTA, pH 8. PCR was performed using the following primers: WT F‐CCAACTGTTATACTCAACCC (GenBank: JN959483.1 13231–13250) and WT R‐GTTTTACTCACGGGCATAGC (JN959483.1 13662–13643), flanking AmelX exon two and yielding a 432 bp PCR product; KO F‐GCATCCTTAGGCTGTATAGC (JN959483.1 13292–13311) and KO R‐GCTTTACGGTATCGCCGCT (with the NeoR cassette inserted), yielding an approximately 700 bp PCR product.

Notably, amelogenin‐null mice exhibited a lower body weight compared to wild‐type mice at all ages and displayed hypoplastic enamel structures on micro‐CT scans (Figure S1), as described earlier [2, 6].

2.3.1. Study Design of Amelogenin‐Null Mice

Twenty‐eight female mice were included in this study. The animals were allocated randomly into experimental groups as follows: (i) 23‐month‐old wild‐type (n = 10) and amelogenin‐null (n = 10) mice were examined using either micro‐CT scans (n = 5 per group) or histological analyses (n = 5 per group); (ii) 12‐month‐old wild‐type (n = 4) and amelogenin‐null (n = 4) mice were examined using histological analyses.

2.3.2. Micro‐CT Analyses

Following euthanasia, the jaws, knees, and segments of the vertebral column were harvested and fixed in 4% paraformaldehyde (PFA) (Santa Cruz Biotechnology). The tissues were then dehydrated in increasing ethanol concentrations. They were scanned using a high‐resolution micro‐CT40 scanner (SCANO Medical, Bassersdof, Switzerland) at a 20 μm resolution, 45 kV energy, and 0.176 mAs. For analysis, two‐dimensional (2D) coronal slices and three‐dimensional (3D) reconstructions of the jaws, knees, and spines were generated using SCANCO software. Image processing was performed with a Gaussian filter (sigma = 0.8 and support = 1), a threshold of 224 for 2D scans, and a threshold of 316 for 3D reconstructions.

2.3.3. Tissue Processing and Histological and Immunohistochemical Analyses

Harvested mouse tissues were fixed in 4% PFA (Santa Cruz Biotechnology) overnight, washed in phosphate‐buffered saline (PBS), and decalcified in 25% formic acid (JTBaker) in citrate buffer (Tivan Biotech) until a soft consistency was achieved. The tissues were then neutralized with 0.25% ammonium hydroxide (Fluka), soaked in a 30% sucrose solution, and embedded in an optimal cutting temperature (OCT) compound (Sakura). Sections were obtained using a cryostat (Lecia Biosystems, Germany).

For histological analysis, tissue sections were stained with (i) hematoxylin and eosin (H&E) (Mayer's hematoxylin, Bio Optica, Milano, Italy; Eosin, Surgipath Medical Industries Inc., Richmond, IL, USA), (ii) toluidine blue (Millipore Sigma, USA), and (iii) safranin O/fast green (Millipore Sigma, USA). Indirect immunohistochemistry was performed using primary antibodies directed against type II collagen (AM26374PU‐N, ORIGENE) or matrix metalloproteinase 13 (MMP‐13; TA321485, ORIGENE). The signal was developed using a Histostain‐SP kit (Invitrogen) according to the manufacturer's instructions. The slides were then mounted (Fluoromount, DBS, Pleasanton, CA, USA) and examined using an Axioskop microscope (Zeiss, Göttingen, Germany). Images were captured using a C10 ProgRes camera (Jenoptik, Jena, Germany).

2.4. Induction of OA in a Rat Model and Treatment

Eight‐week‐old Wistar (RccHan:WIST) female rats, with a mean weight of 215 ± 6.8 g, were treated as described below. All surgical procedures were performed by an orthopedic surgeon who was blinded to the treatment applied.

2.4.1. Destabilization of the Medial Meniscus Model (DMM)

The rats were anesthetized with isoflurane inhalation and received subcutaneous tramadol (Grunenthal) for analgesia prior to surgery. Under aseptic conditions, using the medial parapatellar approach, the right knee joint patella was everted to expose the tibial articular surface and the menisci. Transection of the medial meniscotibial ligament (MTL) was performed to displace the medial meniscus medially. The arthrotomy and skin were sutured using interrupted 5–0 vicryl and 5–0 nylon sutures, respectively. Postoperatively, the rats received pain relief medication (PO dipyrone; Teva) for 3 days and were monitored twice a week until euthanasia. Clinical evaluation included assessment of the Grimace scale, fur appearance, gait, and weight loss.

2.4.2. Preparation of rHAM⁺

The recombinant human amelogenin protein (rHAM⁺) was produced according to Taylor et al. [12] and prepared prior to use according to Helwa‐Shalom et al. [16].

2.4.3. Meniscal Restabilization

Eight weeks following DMM induction, the rats were anesthetized, and the patella was again everted to expose the joint as described above. To halt the progression of osteoarthritic damage, the medial meniscus was restabilized by suturing it to the anterior edge of the medial tibial plateau using a 5–0 nylon suture [19]. The arthrotomy was sutured using interrupted 5–0 vicryl sutures. Before complete closure of the joint capsule, 7 μL of 0.5 mg/mL rHAM⁺ dissolved in its carrier, 2.25% propylene glycol alginate (PGA), or 7 μL of 2.25% PGA carrier alone was applied to the joint. The joint capsule was sealed with a suture, and the skin was closed. Pain relief medication and animal monitoring were conducted as described above.

2.4.4. Rat Knee Experimental Study Design

Forty‐two female rats were included in this study and randomly allocated as follows: (i) Eight rats were evaluated following the 8‐week induction of OA. An additional 30 rats were further allocated into experimental groups, which were treated with 0.5 mg/mL rHAM⁺ dissolved in 2.25% PGA (15 rats), or a control group, treated with 2.25% PGA carrier alone (15 rats). These two groups were further subdivided into the following groups: (ii) 5 days after meniscal restabilization surgery (n = 4 per group); (iii) 2 weeks after meniscal restabilization surgery (n = 3 per group); and (iv) 24 weeks after meniscal restabilization surgery (n = 8 per group). The severity of OA changes was assessed by magnetic resonance imaging (MRI) (groups i and iv), as well as by histological and immunohistochemical analyses (all groups). (v) Four untreated rats were evaluated using MRI, histological, and immunohistochemical analyses.

2.4.5. MRI Analyses

MRI analyses were performed on the knees of the rats 8 weeks after the induction of OA, without additional treatments, and 24 weeks after treatment with 0.5 mg/mL rHAM⁺ in PGA or PGA alone. MRI images were captured using a 7T 24 cm bore cryogen‐free MR scanner based on proprietary dry magnet technology (MR Solutions, Guildford, UK) using a quadrature transmit/receive bird cage volume coil (20 mm inner diameter, 18 mm length). During the MRI, the rat knee was immersed in Fomblin‐YR (ChemCruz) and positioned with the limb's axis parallel to the coil's center. Ten axial T1‐weighted (T1W) spin–echo slices were acquired for orientation purposes (repetition time = 1020 milliseconds, echo time = 11 milliseconds, FOV = 15 mm × 15 mm, 256 × 256, slice thickness = 0.5 mm, four averages). Using these axial slices as references, nine coronal T1W slices were then acquired through the knee joint (repetition time = 3000 milliseconds, echo time = 11 milliseconds, FOV = 20 mm × 20 mm, 256 × 256, slice thickness = 0.4 mm, four averages). The MRI results were analyzed by an orthopedic surgeon and a researcher, both of whom were blinded to the treatment applied, using VivoQuant 4.0 software from Invicro. The quantification of the MRI images was performed according to the whole‐organ MRI score (WORMS) system [20], with higher scores representing greater degrees of joint degeneration.

2.4.6. Tissue Processing and Histological and Immunohistochemical Analyses

Harvested rat tissues were processed and stained according to the methods described in Helwa‐Shalom et al. [16], with the following modifications and additions: (i) decalcification was performed using the MoL‐DECALCIFIER (MILESTONE) solution; (ii) immunohistochemical staining for MMP‐13 was performed using the primary antibody (PA5‐27242 Invitrogen) and secondary antibody (ZUC032 Zytomed); and (iii) immunohistochemical staining for amelogenin was performed using a custom designed rabbit polyclonal antibody (Sigma‐Aldrich) directed against the middle hydrophobic part of the protein as the primary antibody. The procedure was performed using the Rabbit specific HRP/DAB detection IHC detection kit (ab236469 Abcam), according to the manufacturer's instructions.

Semiquantitative analysis of H&E‐stained tissues was performed using the modified Mankin classification of osteoarthritis [21], with a scale ranging from 0 (normal tissue) to 18 (severe OA phenotype). In addition, semiquantitative analysis based on the three zonal depth ratio of the OARSI histopathology initiative [22] was performed, where a lower score indicates a higher degree of cartilage degradation.

Quantitative analyses of histological and immunohistochemical images were performed using QuPath 0.5.1 software [23]. Quantification was performed on the entire cartilage or separately for the superficial layer, extending from the surface to the central part of the cartilage, and the deeper zone, from the middle to the tidemark. Cartilage width was measured in three different points along each sample. MMP‐13‐positive cells were evaluated on two slides from each rat. CD105‐positive and negative cells were evaluated on three slides from each rat. Positively (brown) and/or negatively (purple) stained cells from the medial tibial plateau were counted and averaged for each rat.

2.5. Statistics

Statistical analyses were conducted using GraphPad Prism 9 software (GraphPad Software, USA; RRID:SCR_002798). Normality was assessed using the Shapiro–Wilk test. Data regarding mice experiments were compared using an unpaired Student's t‐test. For rat data, statistical comparisons were performed as follows: MRI analyses were evaluated using one‐way analysis of variance (ANOVA) followed by Tukey's post hoc test for multiple comparisons; semiquantitative histological scoring and quantification of positively/negatively stained cells were analyzed using the Mann–Whitney U test; histological assessments were analyzed using an unpaired Student's t‐test or ANOVA test, as detailed below. Data are presented as the mean value and standard error of the mean (SEM). A p‐value of ≤ 0.05 was considered statistically significant.

3. Results

3.1. Twenty‐Three‐Month‐Old Amelogenin‐Null Mice Display Osteoarthritic Changes in the Knee and Spine Joints

The only phenotypes described in amelogenin‐null mice are a lack of normal prism structure (hypoplastic enamel) and reduced body weight throughout life [2, 3, 4, 5, 6]. We evaluated the knee and spine joints of aging amelogenin‐null and wild‐type mice. Micro‐CT analyses of the knees of 23‐month‐old amelogenin‐null mice revealed advanced osteoarthritic changes, including subchondral sclerosis, significant joint space narrowing, and osteophyte formation (Figure 2C,D). These phenotypes were not observed in age‐matched wild‐type mice (Figure 2A,B).

FIGURE 2.

Micro‐computed tomography (micro‐CT) scans of 23‐month‐old amelogenin‐null mice knees compared with wild‐type knees. 2D coronal slices (A, C) and 3D reconstructions (B, D) of micro‐CT representative images of wild‐type (WT) (A, B) or amelogenin‐null (C, D) mice. The knee joints of amelogenin‐null mice displayed advanced osteoarthritic changes, including narrowing of the medial articular joint space with subchondral sclerosis (blue arrow) and large osteophytes around the knee (yellow arrows).

Histological analyses of the knee joints of 23‐month‐old amelogenin‐null mice revealed advanced tissue destruction, whereas the knee joints of 23‐month‐old wild‐type mice presented only minor age‐related matrix loss (Figures 3 and 4). Knee tissue destruction in aging amelogenin‐null mice included profound matrix loss, destruction of the hyaline cartilage, exposure of the calcified cartilage and the subchondral bone, and the presence of ectopic osteophytes (Figures 3C–E and 4B). Semiquantitative analyses of the osteoarthritic changes confirmed a greater extent of damage in the amelogenin‐null mice compared to wild‐type mice (mean 15.9 ± SE 1.1 vs. mean 8.2 ± SE 2.1, respectively, p = 0.0065) (Figure 4C).

FIGURE 3.

Histological evaluation of the knee joints of 23‐month‐old amelogenin‐null mice compared with wild‐type mice. Hematoxylin–eosin (A, C) and toluidine blue (B, D, E) histological analyses of thin sections obtained from the knee joints of wild‐type (WT) (A, B) or amelogenin‐null mice (C, E). Minor cartilage changes were observed in wild‐type mice, representing the normal aging process (A, B). Enlarged pictures (A–D) were taken from the regions marked with black frames. The amelogenin‐null joints displayed a large osteophyte at the edge of the joint (yellow arrow) (E). Significant cartilage destruction, including matrix loss exposing the calcified cartilage (red arrowheads), represents an osteoarthritic process (C, D). CZ, calcified zone; F, femur; MZ, middle zone; SB: subchondral bone; SZ, superficial zone; T, tibia. Scale bars: 250 μm for overall images; 62.5 μm for enlarged images.

FIGURE 4.

Safranin O‐Fast green staining of knee joints from five 23‐month‐old amelogenin‐null mice compared with five wild‐type mice. Representative thin sections obtained from the knee joints of wild‐type (WT) (A) or amelogenin‐null mice (B) are displayed. The wild‐type knee joints showed minor cartilage destruction, indicating the normal aging process, except for one knee where matrix loss partially exposed the calcified cartilage (red arrowheads). The amelogenin‐null joints displayed severe cartilage destruction, including exposure of the subchondral bone (black arrowheads). F, femur; T, tibia; scale bars, 250 μm. Grading of osteoarthritic changes in the medial tibial plateau of knee joints from 23‐month‐old amelogenin‐null mice (n = 5) compared with wild‐type mice (n = 5) was performed using modified Mankin classification for osteoarthritis (C). The data are presented as the means (standard errors of the means). **p ≤ 0.01, Unpaired student's t‐test.

Similar osteoarthritic changes were detected in the spinal column joints of amelogenin‐null mice. These included irregular disc surfaces, decreased intervertebral disc height, and osteophyte outgrowth (Figure 5D–F). In contrast, wild‐type mice exhibited normal spinal column structure (Figure 5A–C). Therefore, we concluded that the osteoarthritic changes in aged amelogenin‐null mice were not confined to a single joint.

FIGURE 5.

Micro‐CT analyses of the spine columns of 23‐month‐old amelogenin‐null mice compared with those of wild‐type mice. 2D coronal slices (A, D) and 3D reconstructions (B, C, E, F) of micro‐CT representative images of wild‐type (WT) (A–C) or amelogenin‐null (D–F) mice. The amelogenin‐null joints presented advanced osteoarthritic changes, including irregular end plates, narrowed disc space (pink arrow), and large osteophytes around the discs and the facet joints (yellow arrows). Scale bars, 1 mm.

3.2. Twelve‐Month‐Old Amelogenin‐Null Mice Exhibit Reduced Type II Collagen Levels and Matrix Metalloproteinase‐13 Accumulation Compared to Wild‐Type Mice

We evaluated earlier tissue destruction processes in 12‐month‐old wild‐type and amelogenin‐null mice to correlate such events with the osteoarthritic phenotype of 23‐month‐old amelogenin‐null mice. We assessed the expression of type II collagen, the major protein component of the cartilage extracellular matrix (ECM), and matrix metalloproteinase‐13 (MMP‐13), the primary MMP involved in the cleavage of type II collagen [24]. Immunohistochemical analyses of knee joints from 12‐month‐old amelogenin‐null and wild‐type mice revealed a significant reduction in type II collagen levels and an accumulation of MMP‐13 in amelogenin‐null mice compared to wild‐type controls (Figure 6). Furthermore, hypertrophic chondrocytes were detected in amelogenin‐null knee joint tissues (Figure 6D), further contributing to the loss of cartilage homeostasis and, subsequently, to the development of OA [25].

FIGURE 6.

Immunohistochemical analyses of the knee joints from 12‐month‐old amelogenin‐null and wild‐type mice. Representative images of immunohistochemistry for type II collagen (A, C) and matrix metalloproteinase‐13 (MMP‐13) (B, D) in wild‐type (WT) (A, B) and amelogenin‐null (C, D) mice. Amelogenin‐null mice exhibited reduced type II collagen levels and increased MMP‐13 expression compared to wild‐type mice. These changes were accompanied by cartilage destruction and the presence of hypertrophic chondrocytes (black arrowheads). Enlarged pictures were taken from the regions marked with black frames. Brown staining indicates positive expression of type II collagen or MMP‐13. Scale bars: 250 μm for overall images; 62.5 μm for enlarged images.

3.3. rHAM ⁺ Improves the Induced OA Phenotype in Rats 24 Weeks Posttreatment

The lack of amelogenin in amelogenin‐null mice leads to the development of OA. Next, we evaluated the effect of treating OA with amelogenin in an experimentally induced rat knee model of OA. For this purpose, OA was induced in rat knees using the medial meniscus destabilization (DMM) model. No clinical or behavioral changes were observed in any of the operated rats. Eight weeks post‐induction, OA phenotypes had developed, including cartilage surface irregularities, matrix loss extending into the middle zone, and decreased staining for proteoglycans (Figure 7).

FIGURE 7.

Induction of OA in a Rat Model. Representative images of rat knee joints: Toluidine blue histological staining (A, C) and MRI scans (B, D). Representative normal, unoperated knees (A, B) and operated rat knees 8 weeks after medial meniscus destabilization (C, D). Induction of OA was evident in the operated rats; Cartilage loss (yellow arrow). F, femur; L, lateral meniscus; M, medial meniscus; T, tibia. Scale bar: 100 μm.

Following confirmation of OA induction, the medial meniscus was surgically restabilized, and the knee joint was treated with either 0.5 mg/mL rHAM⁺ dissolved in PGA carrier (experimental group) or with the carrier alone (control group).

MRI scans revealed partial cartilage repair at the medial plateau, signs of meniscal surgical repair, and minimal bone marrow abnormalities, 24 weeks after treatment with 0.5 mg/mL rHAM⁺. In contrast, knees treated with PGA carrier alone revealed multiple areas of full‐thickness cartilage loss, cysts, marrow abnormalities, and subchondral bone attrition (Figure 8B,C). Semiquantitative analyses using the WORMS system [20] showed a trend toward cartilage repair in the medial compartment following treatment with 0.5 mg/mL rHAM⁺ compared to treatment with PGA (8.75 ± (SE 1.49) and 12.38 ± (SE 0.63), respectively; p = 0.110, a higher score represents greater degree of joint degeneration). A similar trend of medial cartilage repair following treatment with rHAM⁺ was also evident compared to the OA‐damaged state observed at 8 weeks post‐DMM (Figure 8A) (11.88 ± (SE 1.32), p = 0.186). In contrast, treatment with PGA revealed no improvement of the medial plateau compared to the OA phenotype observed at 8 weeks post‐DMM (p = 0.954), indicating sustained OA (Figure 8D).

FIGURE 8.

Representative Magnetic Resonance Imaging (MRI) scans. Eight weeks after induction of experimental OA (DMM surgery) in a rat knee model (A) and 24 weeks followeing destabilization and treatment with 0.5 mg/mL rHAM⁺ dissolved in a 2.25% PGA carrier (experimental) (B), or with 2.25% PGA carrier alone (control) (C). F, femur; LM, lateral meniscus; MM, medial meniscus; T, tibia; Lines mark the edges of the cartilage defect; Arrowhead, mark the lateral cartilage. (D, E) Semiquantitative scoring of joint degeneration according to the whole‐organ magnetic resonance imaging score (WORMS) [20]: Semi quantification of the medial compartment (D) and the lateral compartment (E) (n = 8 per group). A lower score indicates fewer signs of OA. Less degeneration of the joint was observed following treatment with 0.5 mg/mL rHAM⁺ compared to the PGA carrier. Data are presented as the mean (standard error of the mean); *p ≤ 0.05; **p ≤ 0.01, ANOVA with post hoc Tukey's analysis for multiple comparisons.

Significant repair was demonstrated in the lateral compartment after the treatment with 0.5 mg/mL rHAM⁺ compared to treatment with PGA carrier (7.12 ± (SE 1.43) and 11.00 ± (SE 0.76), respectively; p = 0.0447). The lateral compartment remained relatively stable compared to the OA phenotype post‐DMM surgery (5.88 ± (SE 0.88), p = 0.686), indicating that amelogenin protected from the deterioration of the already existing damage in the lateral joint compartment. However, significant deterioration of the lateral compartment was observed after treatment with PGA (p = 0.007), reflecting the progression of OA beyond the initially mildly affected area (Figure 8E).

Histological and immunohistochemical analyses 24 weeks posttreatment revealed thicker articular cartilage after treatment with rHAM⁺ (mean 243.4 μm [SE ± 13.72 μm]) with mainly smooth, uninterrupted surfaces and only mild irregularities (Figure 9A,C, Figure S2A). In contrast, the PGA‐treated group exhibited significantly thinner cartilage (mean 191.2 μm [SE ± 22.59 μm], p = 0.034) with large fissures extending into the deep zone (Figure 9B,D, Figure S2B). Type II collagen, a major component of the extracellular matrix (ECM), was more extensively expressed throughout the cartilaginous layers of the medial compartment following treatment with 0.5 mg/mL rHAM⁺, compared to a milder expression following treatment with PGA (rHAM⁺: mean intensity unit 0.73 [SE ± 0.06]; PGA—0.62 [SE ± 0.02], p = 0.04) (Figure 9E,F).

FIGURE 9.

Histological and immunohistochemical evaluation of induced osteoarthritic knee joints in a rat model 24 weeks after treatment with 0.5 mg/mL rHAM⁺ dissolved in the PGA carrier (experimental) or only with PGA carrier (control). Rats were treated with either 0.5 mg/mL rHAM⁺ dissolved in the 2.25% PGA carrier (experimental, A, C, E, G, I) or with 2.25% PGA carrier alone (control, B, D, F, H, J). (A, B) Hematoxylin–eosin (H&E) histological staining. (C, D) Toluidine blue histological staining and measurement of cartilage thickness. (E, F) Immunohistochemical staining and quantification of type II collagen. (G, H) Immunohistochemical staining and quantification of type X collagen at the surface zone (SZ) or the deeper zone (DZ). (I, J) Immunohistochemical staining and quantification of MMP‐13. Enlarged black‐framed H&E images were taken from the regions marked with dashed black frames in Figure S2. Enlarged red‐framed images were taken from the regions marked with dashed red frames in (I, J). Scale bar: 250 μm; Scale bar of enlarged red‐framed images: 100 μm. Black arrowhead, fissure extending into the deep cartilaginous zone; F, femur; T, tibia. (K–M) Semiquantitative modified Mankin classification of osteoarthritis [21] of the medial tibial plateau (K), the lateral‐tibial plateau (L), and whole‐joint repair (M) (n = 8 per group). A lower score indicates fewer signs of OA. (N) Semiquantification of zonal depth ratio of lesions at three points along the cartilage surface, the OARSI histopathology initiative [22] (n = 8 per group). A lower score indicates higher degree of cartilage degradation. A more advanced osteoarthritic process was observed following treatment with the PGA carrier compared to rHAM⁺. The data are presented as the means (standard errors of the means). *p ≤ 0.05; **p ≤ 0.01; Mann–Whitney U test.

Expression of the well‐established markers of hypertrophic chondrocytes, type X collagen, and MMP‐13 [26] was also evaluated (Figure 9G–J). The expression level of type X collagen was similar after treatment with 0.5 mg/mL rHAM⁺ or PGA in the deeper layers of tibial cartilage, where it is typically expressed under normal physiological conditions [27] (rHAM⁺: mean intensity unit 0.45 [SE ± 0.046]; PGA 0.39 [SE ± 0.01], p = 0.26). However, in the surface zone of the tibial articular cartilage, treatment with PGA resulted in a significantly higher type X collagen expression compared to treatment with rHAM⁺ (PGA: mean intensity unit 0.54 [SE ± 0.02]; rHAM⁺ − 0.43 [SE ± 0.04], p = 0.034) (Figure 9G,H), demonstrating elevated OA [27] and compared to its expression in the deeper layers of the cartilage (p = 0.006). Contrary to this, after treatment with rHAM⁺, a very mild, insignificant decrease in type X collagen expression was demonstrated in the surface zone (p = 0.717) (Figure 9G,H).

Treatment with PGA also demonstrated a trend toward increased MMP‐13 expression in the medial tibial articular cartilage compared to treatment with rHAM⁺ (PGA: mean intensity unit 0.47 [SE ± 0.03]; rHAM⁺ − 0.41 [SE ± 0.04], p = 0.104) (Figure 9I,J).

The severity of rat knee OA 24 weeks posttreatment with 0.5 mg/mL rHAM⁺ compared to PGA carrier was assessed using a semiquantitative analysis based on modified Mankin score and on OARSI criteria [21, 22]. Significantly milder OA phenotypes were obtained following treatment with rHAM⁺ compared to PGA. The lateral plateau of the rHAM⁺‐treated knees exhibited fewer characteristics of OA than the medial plateau. Significantly higher and similar scores were observed in both the lateral and medial compartments following treatment with PGA, indicating more severe OA phenotypes (medial compartment: rHAM⁺ mean score 7.7 [SE ± 1.4] vs. PGA mean score 10.3 [SE ± 1.2], p = 0.030; lateral compartment: rHAM⁺ mean score 3.7 [SE ± 1.0] vs. PGA mean score 10.0 [SE ± 2.0], p = 0.005) (Figure 9K,L).

Semiquantitative evaluation of the entire knee joint also revealed significantly fewer characteristics of OA after treatment with 0.5 mg/mL rHAM⁺ compared to PGA (rHAM⁺ mean score 5.2 (SE ± 1.1) vs. PGA mean score 9.1 (SE ± 1.2); p = 0.02) (Figure 9M). Furthermore, the zonal depth ratio of the lesions, according to OARSI criteria, reflecting the degree of cartilage degeneration, was significantly worse following treatment with PGA compared to treatment with 0.5 mg/mL rHAM⁺ (PGA mean ratio 0.65 (SE ± 0.14) vs. rHAM⁺ mean ration 0.93 (SE ± 0.03); p = 0.01) (Figure 9N).

These results highlight the protective and reparative effects of applied amelogenin, as it not only inhibited OA progression but also facilitated healing of the medial compartment, the most affected region in this model.

Therefore, we tested whether exogenously added amelogenin (rHAM⁺) induced long‐term elevation of expression of endogenous amelogenin, as a mean to exert its extended protective effect. Untreated (normal) rat joints expressed lower, though non‐significant, levels of endogenous amelogenin compared to all tested post‐DMM surgery conditions: only induction of DMM and 24 weeks post treatment with either rHAM⁺ or PGA (Normal, mean intensity unit 0.36 (SE ± 0.03); DMM—0.44 (SE ± 0.04); PGA—0.44 (SE ± 0.02) p = 0.059 for both groups; rHAM⁺‐ 0.41 (SE ± 0.01), p = 0.246) (Figure S3).

3.4. Increased Type II Collagen Expression and Reduced MMP‐13 Levels Two Weeks After Treatment With rHAM ⁺ Compared to PGA Control

To evaluate the progression of tissue degradation, knee joint sections were immunohistochemically stained for type II collagen and MMP‐13 two weeks after treatment with either 0.5 mg/mL rHAM⁺ in PGA carrier or with PGA carrier alone. Following treatment with rHAM⁺, type II collagen expression was highly pronounced in the superficial zone, with lower but widespread expression throughout the cartilage layers, in the extracellular matrix (ECM) and in hypertrophic chondrocytes (Figure 10A). However, PGA‐treated joints exhibited a noncontinuous, high expression of type II collagen in the superficial zone, with weaker staining in the deeper cartilage layers (Figure 10C). Knee joints treated with PGA exhibited increased MMP‐13 expression. In contrast, joints treated with rHAM⁺ presented significantly lower MMP‐13 levels, particularly in the middle zone of the medial tibial cartilage, demonstrating less aggravated OA progression compared to treatment with PGA (Figure 10B,D). Quantitative analysis revealed significantly fewer MMP‐13‐positive cells in the medial tibial plateau after treatment with rHAM⁺ compared to PGA alone (145.3 cells/rat (SE ± 19.4) and 241 cells/rat (SE ± 10.9), respectively; p = 0.05) (Figure 10E).

FIGURE 10.

Immunohistochemical evaluation of the medial tibial plateau of experimentally induced OA in rats 2 weeks after treatment with 0.5 mg/mL rHAM⁺ dissolved in PGA (experimental) or with PGA carrier alone (control). Type II collagen staining (A, C), or MMP‐13 staining (B, D), following treatment with 0.5 mg/mL rHAM⁺ dissolved in PGA (A, B) or PGA carrier alone (C, D); Enlarged images were taken from the regions marked with dashed frames. Brown staining: Positive for type II collagen or MMP‐13; Purple staining: No staining (ECM) or cell nuclei. Scale bar: 250 μm; Scale bar of enlarged black‐framed images: 50 μm; Scale bar of enlarged red‐framed images: 10 μm. The average number of MMP‐13‐positive cells per rat in the medial tibial plateau (E). A decreased amount of MMP‐13 were presented at the knee joint following treatment with 0.5 mg/mL rHAM⁺ compared to PGA control. *p ≤ 0.05; Mann–Whitney U test.

3.5. Treatment With rHAM ⁺ Facilitates Recruitment of the Mesenchymal Stem Cells Marker CD105

We have previously shown rHAM⁺‐induced recruitment of MSCs toward an injured or diseased joint [13, 14, 16]. We thus evaluated the ability of rHAM⁺ to recruit CD105‐positive mesenchymal stem cells to osteoarthritic rat joints 5 days posttreatment. A trend toward recruitment into the medial tibial cartilage was observed following application of rHAM⁺ compared to PGA carrier (32.8 cells in the medial tibial plateau (SE ± 4.8) and 23.7 (SE ± 3.4), respectively; p = 0.057). The CD105‐positive cells were mainly localized at the edges and the middle superficial region of the medial tibial cartilage, which may indicate recruitment from adjacent joint tissues (Figure 11).

FIGURE 11.

Immunohistochemical evaluation of the medial tibial plateau of experimentally induced OA in rats 5 days after treatment with 0.5 mg/mL rHAM⁺ dissolved in PGA (experimental) or with PGA carrier alone (control). (A, B) CD105 staining following treatment with 0.5 mg/mL rHAM⁺ dissolved in PGA (A) or PGA carrier alone (B); Enlarged images were taken from the regions marked with dashed frames. Brown staining: Positive for CD105; Purple staining: No staining (ECM) or cell nuclei. Scale bar: 200 μm; Scale bar of enlarged framed images: 50 μm; (C) The percentage of CD105‐positive cells per rat in the medial tibial plateau. A trend toward increased level of CD105 was presented at the knee joint following treatment with 0.5 mg/mL rHAM⁺ compared to PGA.

4. Discussion

The amelogenin knockout mouse model has been available since 2001 [2], with published phenotypes limited to defects in tooth‐related structures, resembling the X‐linked form of amelogenesis imperfecta (AI) in humans, along with reduced body weight throughout life [2, 3, 4, 5, 6]. However, no previous studies have investigated the potential skeletal phenotypes in amelogenin‐null mice. In the current study, we found that 12‐month‐old amelogenin‐null mice presented mild OA phenotypes, including cartilage destruction, hypertrophic chondrocytes, reduced type II collagen levels, and accumulation of MMP‐13 in the knee joint. OA progression continued in amelogenin‐null mice, and at the age of 23 months, the amelogenin‐null mice presented phenotypic traits that are considered hallmarks of OA. The knee joints displayed destruction of the hyaline cartilage and subchondral bone, joint space narrowing, subchondral sclerosis, cyst development, and periarticular ectopic osteophytes. Additionally, their spinal column displayed ectopic osteophytes and decreased disc space height, as well as an irregular disc surface. In contrast, wild‐type mice of the same age exhibited only mild, age‐related joint and spinal changes. Amelogenin is expressed in healthy bone and cartilage tissues [11]. Therefore, the depletion of amelogenin from joint and spine tissues may contribute to the development of OA phenotypes observed in aged amelogenin‐null mice. Interestingly, a syndromic form of autosomal AI is caused by mutations in latent transforming growth factor‐beta‐binding protein 3 (LTBP3). Ltbp3‐null mice develop osteosclerosis and osteoarthritis between 6 and 9 months of age. A relationship between amelogenin and transforming growth factor‐beta1 (TGF‐β1) has been previously described in enamel formation, where activated TGF‐β1 binds to major amelogenin cleavage products to maintain their activity. The amelogenin‐TGF‐β1 complex subsequently interacts with TGFBR1, initiating TGF‐β1 signaling.

In light of these results, it would be interesting to evaluate the skeletal structure of aged X‐linked AI patients for OA phenotypes and other skeletal abnormalities. A comparative analysis between affected and unaffected males, as well as between heterozygous (carrier) and homozygous (noncarrier) females within the same family could provide further insights.

To further study the connection between amelogenin and OA, we switched from the mouse amelogenin‐null model to a rat‐induced OA model for consistency, as most experiments involving the application of rHAM⁺ were conducted in rats.

Our previous studies have shown that the application of amelogenin promotes healing in acute skeletal injuries, such as torn ligaments, and induced osteochondral defects in rat knees [14, 15, 16]. Here, we have demonstrated, utilizing MRI scans and histological analyses, that a single application of 0.5 mg/mL rHAM⁺ inhibited the progression of OA and even partially healed the joint, compared to the initial damage and to the OA phenotypes observed after treatment with the PGA control. Following rHAM⁺ treatment, significantly thicker articular cartilage expressing high levels of type II collagen was observed, in contrast to control rats treated with the PGA carrier alone. Additionally, tibial cartilage displayed almost complete repair, characterized by proper cell morphology, a well‐organized extracellular matrix, and continuous integration with the surrounding tissue. Moreover, MRI analyses suggested that treatment with amelogenin protected not only the OA‐induced compartment but also preserved the integrity of the collateral side of the joint. In contrast, treatment with PGA demonstrated severe OA phenotypes, further exacerbating damage in the lateral compartment by 87.4%.

We revealed a trend toward increased expression of endogenous amelogenin after induction of OA (regardless of additional treatments), compared to normal, untreated knee joint. This finding is in line with Mitsiadis et al. [28], who demonstrated elevated expression of the amelogenin in the lesion site of dental injuries. We agree with their proposal that amelogenin may act as a reparative signal [28]. However, its' expression level in injured or diseased tissue is insufficient for effective action, specifically, as we demonstrated here, for prevention of development and progression of OA. Hence a need for exogenous application of amelogenin is evident.

The expression of endogenous amelogenin 24 weeks after treatment with rHAM⁺ was only mildly and insignificantly elevated compared to normal (untreated) knee, which might indicate the need for continuous release of amelogenin at the injured site to prevent the OA phenotype.

OA is characterized by imbalanced homeostasis between cartilage anabolic and catabolic processes, resulting in irreversible structural and functional degradation of joint tissues. MMP‐13 is the main catabolic protease that cleaves cartilaginous type II collagen, making it one of the main factors involved in aggravating OA [29]. The levels of MMP‐13 were elevated in 12‐month‐old amelogenin‐null mice and in OA‐induced rats 2 and 24 weeks after treatment with PGA, demonstrating chronic degenerative processes in the control group. Elevated levels of MMP‐13 were accompanied by reduced expression of type II collagen, as described above. Twenty‐four weeks after treatment with rHAM⁺, the levels of the hypertrophic chondrocyte marker, type X collagen [26], were significantly reduced in the superficial layer relative to PGA control, demonstrating milder OA [27].

Endogenous MSCs reside in articular cartilage, and their number increases during OA. In spite of this, the responsiveness of the endogenous MSC population during OA is impaired [30]. Therefore, in recent years, the application of exogenous MSCs, derived from various tissues, has emerged as a potential treatment for OA, alleviating pain and improving function [31, 32]. Exogenous CD105‐positive MSCs migrated toward the osteoarthritic joints [33, 34] and enhanced type II collagen synthesis while reducing MMP‐13 expression in chondrocytes [35]. However, the long‐term efficacy and safety of such treatments are currently being studied [36]. Previously, we demonstrated amelogenin‐induced paracrine recruitment of cells expressing mesenchymal stem cell markers to the injured site four days after injury in various rat models [13, 14, 16]. Therefore, we evaluated the ability of rHAM⁺ to recruit such cell populations into the damaged zone of induced‐osteoarthritic rat joint and demonstrated a substantial increase in CD105‐positive cells, mainly at the edges of the cartilage and the central superficial zone. The origin, fate, and functional role of these cells, whether recruited paracrinally, expanded in situ, or derived from resident populations, remain to be determined.

Numerous studies have suggested that amelogenin might act as a structural agent, resembling its well‐documented activity during enamel matrix formation, as it self‐assembles into nanospheric structures both in vitro and in vivo [1, 3, 4]. These nanospheres might alter the surface properties of hyaline cartilage or synovial fluid, promoting cell adhesion and recruitment, or may even act as a lubricant due to their hydrophobic nature. Additional possible mechanism(s) have been proposed for soluble extracellular amelogenin. Several cell surface receptors have been proposed for amelogenin, including lysosomal‐associated membrane protein‐1 (LAMP‐1), CD63, and glucose‐regulated protein 78 (GRP78) [1, 37]. Amelogenin may induce its protective/healing effect by binding to one or more of these cellular receptors and initiating an unknown second messenger response. Alternatively, amelogenin bound to these receptors was shown to be rapidly internalized into vesicles, which subsequently localize to the perinuclear region of the cell, likely indicating degradation by lysosomes and late endosomes [1, 37, 38, 39, 40]. Amelogenin peptides might initiate a healing effect [41, 42] however, further studies are needed to decipher the involvement of such amelogenin peptides in alleviating OA and the underlying mechanism(s).

Owing to the degenerative nature of OA, it remains unclear whether a single application of rHAM⁺ could halt disease progression for an extended period. Additional doses at certain intervals or a continuous delivery system may be required for long‐term effects. In conclusion, our study demonstrated that a lack of amelogenin led to the development of osteoarthritic phenotypes, whereas a single application of rHAM⁺ inhibited further deterioration and also partially healed the osteoarthritic phenotypes. These findings suggest that amelogenin has potential as an osteoarthritis‐modifying drug.

Author Contributions

O.H.S.: methodology, formal analysis, investigation, data curation, writing‐original draft preparation, visualization; Y.K.S.: methodology, formal analysis, investigation, data curation, visualization; S.I.M.: investigation; K.G.: investigation, visualization; D.S.: investigation; S.S.: investigation, visualization; C.Y.: investigation, visualization; Y.N.G.: writing‐review and editing; S.B.: writing‐review and editing; D.D.: conceptualization, supervision, funding acquisition; A.B.: conceptualization, resources, writing‐original draft preparation, visualization, supervision, project administration; H.N.: conceptualization, validation, formal analysis, investigation, resources, data curation, writing‐original draft preparation, visualization, supervision, project administration; A.H.: conceptualization, formal analysis, investigation, writing‐original draft preparation, supervision, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1. Hypotrophic enamel structure of 23‐month‐old amelogenin‐null mice compared to wild‐type. Representative images of 3D reconstructions of micro‐CT images of wild‐type (WT) (A) or amelogenin‐null (B) mice. Yellow arrow, enamel structure. Scale bar, 1 mm.

Figure S2. Hematoxylin–eosin (H&E) staining of induced osteoarthritic knee joints in a rat model 24 weeks after treatment with 0.5 mg/mL rHAM⁺ dissolved in the PGA carrier (experimental, A) or with only the PGA carrier (control, B). Enlarged pictures were taken from the region marked with dashed frames and are shown in Figure 9 A, B. Scale bar: 1000 μm; M, medial meniscus; L, lateral meniscus; F, femur; T, tibia.

Figure S3. Immunohistochemical evaluation of endogenous amelogenin in normal and induced osteoarthritic knee joints in a rat model. (A) Untreated rats, or (B‐D) 8 weeks after experimental OA induction (DMM surgery) in a rat knee model (B), and 24 weeks after the induction of OA followed by destabilization and treatment with 0.5 mg/mL rHAM⁺ dissolved in a 2.25% PGA carrier (experimental) (C), or with 2.25% PGA carrier alone (control) (D). Tissues were immunohistochemically stained for endogenous amelogenin. Scale bar: 100 μm; (E) Quantification of the levels of endogenous amelogenin.

Data S1

Helwa‐Shalom O., Kahlon‐Suki Y., Markowitz S. I., et al., “Amelogenin Null Mice Develop Osteoarthritis, While Its Application Mitigates Disease Phenotypes in a Rat Model,” The FASEB Journal 39, no. 14 (2025): e70838, 10.1096/fj.202500827R.

Data Availability Statement

The data that support the findings of this study are included in the article.