Cartilage-specific knockout of miRNA-128a expression normalizes the expression of circadian clock genes (CCGs) and mitigates the severity of osteoarthritis

Abstract

Background

Micro-ribonucleic acids (miRNAs) are involved in osteoarthritis (OA) pathogenesis and clock-controlled genes (CCGs) regulation. However, the interaction between miRNAs and CCGs remains unclear.

Methods

Human OA samples were used to assess CCGs expression. Cartilage-specific miR-128a knockout mouse model was established to investigate miR-128a′s role in OA pathogenesis. Destabilization of the medial meniscus (DMM) model was employed to simulate OA.

Results

Transcription levels of nuclear receptor subfamily 1 group D member 2 (NR1D2) were lower in both human OA samples and wild-type mice undergoing DMM compared to non-OA counterparts. MiR-128a knockout mice showed reduced disturbances in micro-computed tomographic and kinematic parameters following DMM, as well as less severe histologic cartilage loss. Immunohistochemistry staining revealed a lesser decrease in NR1D2-positive chondrocytes after DMM in miR-128a knockout mice than in wild-type mice. NR1D2 agonist rescued the suppressed expression of cartilage anabolic factors and extracellular matrix deposition caused by miR-128a precursor.

Conclusions

Cartilage-specific miR-128a knockout mice exhibited reduced severity, less disrupted kinematic parameters, and suppressed NR1D2 expression after DMM. NR1D2 enhanced the expression of cartilage anabolic factors and extracellular matrix deposition. These findings highlight the potential of employing miR-128a and CCG-targeted therapy for knee OA.

At a glance commentary

Scientific background on the subject

Micro-ribonucleic acids (miRNAs) play integral roles in both the pathogenesis of osteoarthritis (OA) and the regulation of clock-controlled gene (CCG) expression. However, the precise impact of the interplay between miRNAs and CCGs in OA pathogenesis remains elusive.

What this study adds to the field

Inhibition of cartilage-specific miR-128a resulted in reduced severity, less disrupted kinematic parameters, and suppressed expression of nuclear receptor subfamily 1 group D member 2 (NR1D2) in our OA mouse model. NR1D2, in turn, facilitates cartilage anabolism. Our findings underscore the potential of utilising miR-128a and CCG-targeted therapy for knee osteoarthritis.

Osteoarthritis (OA) is a debilitating disease manifested by the progressive loss of articular cartilage and is a pivotal cause of disability in the aged society [1,2]. OA is a multifactorial disease, and the whole picture of the OA pathogenesis has not been completely delineated [2].

Emerging studies have suggested that non-coding, single-stranded micro-ribonucleic acids (miRNAs) serve as the major player in OA pathogenesis [3]. MiRNAs could mediate the expression of target genes at the post-transcriptional level [4,5]. By base pairing with the seed sequence of target mRNA molecules, the 3′-untranslated region (3′-UTR), miRNAs could perturb the expression of target genes. Via binding to the 3′-UTR of target mRNA molecules, miRNAs could participate in myriads of physiologic processes, such as the regulation of circadian rhythm [6].

Disturbed circadian rhythm has been recognized to be involved in the pathogenesis of OA [7]. Some authors have found that variations in cartilage thickness could be observed in healthy humans throughout the day, and a slow decrease of cartilage occurs over the day before recovering overnight [8]. Circadian alternations are present in the clinical presentation (pain sensation, stiffness, and dexterity) and biomarkers of OA patients [9,10]. Substantial evidence has suggested the presence of functional circadian clock in cartilage tissue capable of driving downstream clock-controlled genes (CCGs). CCGs are involved in the chondrocyte metabolism and homeostasis of cartilage tissues, and are subjective to post-transcriptional and post-translational regulation based on a complicated transcription and translation feedback system [7]. miRNAs could regulate circadian rhythm by tuning the expression of CCGs [11].

Usually highly expressed in the brain, miR-128 is initially known for its function during neurogenesis and cardiac development [12]. Several studies have shown that abnormal expression of miR-128 is relevant to musculoskeletal diseases. The overexpression of miR-128 reduces migration and epithelial-to mesenchymal transition potential of osteosarcoma [13]. Our team has previously shown that miR-128a expression could exacerbate knee OA by repressing chondrocyte autophagy, triggering our interest for the search of the role of miR-128a in OA pathogenesis [14].

Despite the involvement of miRNAs in the OA pathogenesis and in the regulation of CCGs expression, the crosstalk between miRNAs, CCGs, and OA progression is not clear at present. In the present study, we want to explore the potential crosstalk between miR-128a, CCGs, and OA progression. We hypothesize that knockout of miRNA-128a expression normalizes the expression of CCGs and mitigates the severity of osteoarthritis.

Material and methods

Harv2esting human osteoarthritic and non-osteoarthritic cartilage samples

For collection of clinical specimens, ethical approval was obtained from the Institutional Review Board of Chang Gung Memorial Hospital (Approval No 201802125B0). After written informed consent was obtained, 40 patients with end-stage OA of the knee (29 females and 11 males; 69.4 ± 7.3 years) undergoing total knee arthroplasty were recruited. We harvested cartilage tissues from both severely osteoarthritic regions and regions without gross attrition (e.g., the lateral femoral condyle in varus knee) during total knee arthroplasty to serve as OA and non-OA samples (confirmed by two orthopedic surgeons) [15].

Generation of cartilage-specific miRNA-128a knockout mice

We established cartilage-specific miR-128a knockout (KO) mice to determine the role of miR-128a in the pathogenesis of OA and its interplay with CCGs [Fig. 1]. C57BL/6J mice carrying cartilage-specific Cre recombinase transgene driven by collagen 2α1 promoter (https://www.jax.org/strain/003554) (Col2a1-Cre transgenic mice) were purchased from Jackson lab (Bar Harbor, Maine), and miR-128a promoter loxP-flanked C57BL/6J mice were purchased from the Transgenic Mouse Models Core facility of the National Core Facility for Biopharmaceuticals (Taipei, Taiwan) (http://140.112.133.74/english.htm). Col2a1-Cre transgenic mice and miR-128a promoter loxP-flanked mice were mated together to generate 1st generation progeny. The 1st generation progeny carrying heterozygous loxP-flanked allele and Col2a1-Cre transgene were bred and mated, and the 2^nd generation progeny carrying homozygous loxP-flanked allele and Col2a1-Cre transgene were considered as carrying KO genotype. The miR-128a cartilage-specific KO mice were generated after at least six generations of backcrossing. The mice without Cre recombinase expression and without homozygous miR-128a promoter loxP-flanked allele were consider wild-type mice. Both wild-type and miR-128a KO C57BL/6J mice of both sexes generated by the protocol mentioned above were employed for our studies. For verifying the genotype, genomic DNA from tail tissues in each mouse was harvested and probed by following primers

Fig. 1.

Schematic representation of the generation process of miR-128a cartilage-specific knockout mice.

Forward: 5′-AGGTGAC GTAATTCAGD-3’;

Reverse: 5′-CAATTGCTCATATGGACATGTAC-3′ with polymerase chain reaction (PCR) approaches.

Experimental protocols for laboratory animals were approved by the IACUC of Kaohsiung Chang Gung Memorial Hospital (Nos. 2018121501) and conducted in a specific pathogen-free vivarium.

Generation of osteoarthritis (OA) model by destabilization of medial meniscus (DMM) procedure

DMM procedure was employed to establish the experimental OA model in our study. Eight-week-old male and female mice were anesthetized, and arthrotomy was performed in an aseptic manner. The transection of the medial meniscotibial ligament was performed on left knee of each mouse under a Zeiss surgical microscope [16]. Mice receiving sham operation without transecting the medial meniscotibial ligament were assigned to control group. There was approximately an equal distribution of male and female mice subjected to the DMM procedure or the sham operation.

Micro-computed tomography (micro-CT) analyses

To investigate whether miR-128a signaling was involved in osteophyte formation in mice undergoing DMM procedure, we applied micro-CT imaging technology to assess osteophyte morphology and volume. Mice knee joints are scanned by a SkyScan 1176 micro-CT system (Bruker) to capture 9 μm voxel/slice images. Four hundred transverse images of the region of interest between the articular end and growth plate subjected to reconstruction into three dimensional images using SKYSCAN® CT-Analyser software, following our previous work [14]. Osteophyte volume was quantified using the system's software.

Gait profile analyses

To investigate whether the DMM-induced changes in kinematic parameters would be milder in miR-128a knockout mice, we utilized the Catwalk® system (Noldus Information Technology) to monitor animals' walking patterns and posture. Static gait profiles (footprint area, intensity & contact area) and dynamic gait profile (swing time, distance, speed, stand time), as well as body posture deviation angel are measured with CatWalk software 9.1 and CatWalk XT's Automatic Footprint Classification software, according to the maker's instructions. For assaying animals' mobility, we also have set up a TSI multifunction open field assay system. Animals are put in an area (30 cm × 30 cm × 30 cm) for walking freely for 15 min. Each animal's mobility is monitored. Mobility time and distance were measured automatically by the system.

Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)

Finally, mice were euthanatized, and knee joints were extracted for subsequent studies. The extracted cartilage samples were subjective to reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR) assay, following the protocol of our previous publications [14]. Specimens were homogenized using a Precellys 24® homogenizer and a liquid nitrogen cooling system (Bertin Technologies, Montignyle- Bretonneuz, France). Total RNA was then extracted from RiboPure™ RNA Purification Kit (Thermo Fisher Scientific). After reverse transcription of 1 μg total RNA that was extracted using GoScript™ Reverse Transcription Mix, Oligo(dT), aliquots of RT products were mixed with specific primers (cartilage markers collagen II, aggrecan; circadian regulators NR1D1, NR1D2, BMAL1, Per1, Per2, Per1 and 18S) and Fast SYBR™ Green Master Mix (Applied Biosystems) used for PCR, with reactions performed at 95 °C for 20 s, then 40 cycles at 95 °C for 3 s and 60 °C for 20s. Total microRNA was then extracted from the homogenates using the TaqMan™ MicroRNA Reverse Transcription Kit (Applied Biosystems Inc, Foster City, CA) and further incubated with 2 × TaqMan® PreAmp Master Mix and 10 × MegaplexTM Pre-Amp Primers using an StepOne and StepOnePlus Real-Time PCR Systems (Applied Biosystems Inc, Foster City, CA). The start of the logarithmic amplification of the PCR reactions was computed automatically and interpreted as cycle threshold (Ct). Relative expression was calculated according to Eq. 2^−ΔΔCt, where ΔΔCt = ΔCt_DMM −ΔCt_sham and ΔCt = CtmicroRNA − CtU6, was adapted to quantify changes in microRNA expression in the DMM group. The RT-qPCR assay for human samples from total knee and bipolar hip arthroplasty mentioned above followed the sample protocol.

Histologic analyses

The severity of articular cartilage injury at the proximal tibiae of 10 sections spanning 400 μm was scored using the OARSI (osteoarthritis research society international) score [17]. Histomorphometry of articular tissue was analyzed using a Zeiss microscope and image analysis system. Safranin-O staining of sections was performed according to histochemical protocol (Sigma-Aldrich). Immunostaining was performed using Mouse/Rabbit PolyDetector DAB HRP Brown kits (Bio SB, Inc) and monoclonal antibodies for NR1D2. To analyze the percentage of NR1D2-positive cells relative to the total cells per high-power field, we used the Image Pro Plus 6.0 software, as described in our previous works [18,19]. A total of 32 fields in 10 sections from 8 mice were analyzed.

Assessment of the synthetic capacity of extracellular matrix

Extracellular matrix of cartilage was probed by Alcain blue micromass staining. Micromass of 5 × 10⁵ chondrocytes harvested from the knee joints of seven-day-old male C57BL/6 mice was incubated in DMEM with 10% fetal bovine serum for 7 days [20,21]. To examine the involvement of NR1D2 in miR-128a mediated cartilage metabolism, micromass was pretreated with the NR1D2 agonist SR9009 at a concentration of $1\times {10}^{-4}$ M (EMD Millipore Corporation, MA, USA) for 48 h. The level of Alcain blue in each micromass culture is dissolved by aliquots of 6M guanidine hydrochloride and spectrophotometrically quantified at a wavelength 570 nm.

Statistical analyses

All statistical analyses were performed by employing GraphPad Prism v5.0 (GraphPad Software Inc., San Diego, CA, USA), and all values were shown as the mean ± S.D. Statistical comparisons were performed by Student's t-test or one-way analysis of variance (ANOVA) (two-tail) followed by post-hoc Bonferroni testing. The statistical difference was deemed significant if the p-value was <0.05 [22,23].

Results

During total knee arthroplasty, cartilage tissues were harvested from severe osteoarthritic regions and from regions without gross attrition in the operated joint to serve as human OA and human non-OA samples. Under RT-qPCR assay, the human OA cartilage samples displayed decreased mRNA levels of aggrecan, collagen II, and SOX9 (SRY-box transcription factor 9) than the human non-OA cartilage samples. As for CCGs, human OA samples demonstrated significantly lower NR1D1 (nuclear receptor subfamily 1 group D member 1), lower NR1D2 (nuclear receptor subfamily 1 group D member 2) and lower PER1 (Period Circadian Regulator 1) mRNA levels [Fig. 2].

Fig. 2.

The mRNA expression levels of cartilage anabolic factors and circadian clock genes (CCGs) in human non-osteoarthritis (OA) and OA cartilage samples. Cartilage tissues were harvested during total knee arthroplasty from regions of the operated joint with severe osteoarthritis as well as from regions without significant attrition. These samples were used to represent human OA and human non-OA tissues, respectively. (A) aggrecan; (B) collagen type II; (C) SOX9 (SRY-box transcription factor 9); (D) NR1D1 (nuclear receptor subfamily 1 group D member 1); (E) NR1D2 (nuclear receptor subfamily 1 group D member 2); (F) BMAL1 (basic helix-loop-helix ARNT like 1); (G) PER1 (period circadian regulator 1); (H) PER2 (period circadian regulator 2); and (I) PER3 (period circadian regulator 3) (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001) (n = 40 for non-OA, and n = 40 for OA).

The miR-128a cartilage-specific KO mice and the destabilization of the medial meniscus (DMM) model were established to assess the impact of miR-128a on the pathogenesis of OA and pertinent CCGs expression. All of the subsequent studies were performed 8 weeks after DMM procedures. The average body weight of wild-type and KO mice at 8 weeks was 23.29 g and 21.78 g, respectively (p = 0.007).

micro-CT imaging was employed to demonstrate the radiographic severity of OA 8 weeks after DMM. DMM procedure led to significant increase in osteophyte volume in wild-type mice. The post-DMM bone mineral density (BMD) and post-DMM bone volume/total volume (BV/TV) were all higher in KO mice. The increase in osteophyte volume and osteophyte BV/TV after DMM procedure was significantly less severe in miR-128a knockout mice than in wild-type mice [Fig. 3]. No significant gender-related differences were observed in BMD, BV/TV, trabecular thickness, trabecular number, osteophyte volume, and osteophyte BV/TV when comparing male and female mice with all four groups combined (all p > 0.05).

Fig. 3.

The micro-CT assessments of the OA knees. The osteophytes were marked by red circles and red arrows. (A) The region of interest for micro-CT assessments (marked with red color); (B) Bone mineral density of proximal tibia; (C) Bone volume/total volume of proximal tibia (shown in percentage); (D) Trabecular thickness (shown in millimeter); (E) Trabecular number (shown in reverse millimeter); (F) Representative micro-CT images of the OA knee with medial tibiofemoral compartment osteophytes marked with red circles; (G) Osteophyte bone volume; and (H) Osteophyte bone volume/total volume.

Abbreviations: BMD: bone mineral density; BV/TV: bone volume/total volume; Tb.Th: trabecular thickness; Tb.N: trabecular number. KO: cartilage-specific knockout of miR-128a; WT: wild-type; DMM: destabilization of the medial meniscus (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001) (male/female = 4/4 for WT, male/female = 4/4 for WT + DMM, male/female = 4/4 for KO, and male/female = 4/4 for KO + DMM).

We further investigated whether the DMM-induced OA changes in kinematic parameters would be milder in miR-128a knockout mice. DMM procedures would lead to decreased swing speed, increased swing time, decreased duty cycle, and decreased print area. These changes after DMM procedures were substantially milder in the miR-128a knockout mice than in wild-type mice [Fig. 4]. No significant gender-related differences were observed in swing speed, swing time, duty cycle, print area, run duration, and run average speed when comparing male and female mice with all four groups combined (all p > 0.05).

Fig. 4.

The kinematic parameters of the wild-type and miR-128a knockout mice 8 weeks after DMM procedures. (A) Swing speed (cm/s); (B) Swing (s); (C) Duty cycle (%); (D) Print area (cm²); (E) Run duration (s); and (F) Run average speed (cm/s).

Abbreviations: WT: wild-type; KO: cartilage-specific knockout of miR-128a; DMM: destabilization of the medial meniscus (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001) (male/female = 4/4 for WT, male/female = 4/4 for WT + DMM, male/female = 4/4 for KO, and male/female = 4/4 for KO + DMM).

The wild-type and miR-128a knockout mice undergoing DMM procedure were sacrificed 8 weeks after index procedure, and the cartilage of knees were extracted for subsequent analysis. Under RT-qPCR analysis, the cartilage samples from wild-type mice undergoing DMM procedures showed significantly lower aggrecan, collagen II, and SOX9 expression levels than the wild-type control mice. The decreased expression of aggrecan, collagen II, and SOX9 mRNAs after DMM procedures was significantly milder in miR-128a knockout mice. Among the wild-type mice, the expression levels of miR-128a were significantly higher among the mice undergoing DMM procedures than the mice undergoing sham procedures. Among the mice undergoing sham procedures, the expression levels of miR-128 were effectively suppressed in miR-128a knockout mice. As for CCGs, the cartilage samples from wild-type mice undergoing DMM procedures demonstrated significantly lower NR1D1 and NR1D2 transcription levels than the wild-type controls. The suppressed expression of NR1D2 mRNA after DMM procedures was significantly milder in miR-128a knockout mice [Fig. 5]. There were no significant gender-related differences observed in the transcription levels of miR-128a, aggrecan, collagen II, SOX9, and the selected CCGs when comparing male and female mice with all four groups combined (all p > 0.05).

Fig. 5.

The expression levels of cartilage anabolic factors and CCGs in the wild-type and miR-128a knockout mice undergoing DMM procedures. (A) MiR-128a; (B) aggrecan; (C) collagen type II; (D) SOX9 (SRY-box transcription factor 9); (E) NR1D1 (nuclear receptor subfamily 1 group D member 1); (F) NR1D2 (nuclear receptor subfamily 1 group D member 2); (G) BMAL1 (basic helix-loop-helix ARNT like 1); (H) PER1 (period circadian regulator 1); (I) PER2 (period circadian regulator 2); and (J) PER3 (period circadian regulator 3).

Abbreviations: KO: cartilage-specific knockout of miR-128a; WT: wild-type; DMM: destabilization of the medial meniscus (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001) (male/female = 5/5 for WT, male/female = 5/5 for WT + DMM, male/female = 5/4 for KO, and male/female = 4/4 for KO + DMM).

Safranin-O staining were utilized to illustrate the histologic severity of cartilage loss, and immunohistochemistry (IHC) staining was utilized to demonstrate the local expression level of NR1D2. DMM procedures could lead to apparent cartilage loss and apparent rise in OARSI (osteoarthritis research society international) score in wild-type mice. However, the histologic cartilage loss and the rise in OARSI score after DMM procedures were significantly milder in miR-128a knockout mice. The cartilage layers of both KO and KO + DMM sections appear marginally thicker than from both WT groups [Fig. 6]. In keeping with the results of RT-qPCR assay, DMM procedures led to decreased NR1D2 staining in samples from wild-type mice undergoing DMM procedure than in wild-type control mice, and the decreased NR1D2 staining after DMM procedures was less apparent in samples from miR-128a knockout mice [Fig. 7]. No significant gender-related differences were observed in the OARSI score and the percentage of cells positively stained with NR1D2 when comparing male and female mice with all four groups combined (all p > 0.05).

Fig. 6.

The Safranin-O staining of the cartilage samples from the wild-type and miR-128a knockout mice undergoing DMM procedures. The black and red scale bars represent 1000 and 500 μm, respectively. The regions of magnified images in the overview images of the Safranin O-stained sections were marked with red boxes.

Abbreviations: KO: cartilage-specific knockout of miR-128a; WT: wild-type; DMM: destabilization of the medial meniscus; OARSI: osteoarthritis research society international (∗∗∗p < 0.001) (male/female = 5/5 for WT, male/female = 5/6 for WT + DMM, male/female = 5/4 for KO, and male/female = 4/4 for KO + DMM).

Fig. 7.

The IHC (immunohistochemistry) staining of NR1D2 (nuclear receptor subfamily 1 group D member 2) for the cartilage samples from the wild-type and miR-128a knockout mice undergoing DMM procedures. The black and red scale bars represent 500 and 125 $\mu$ m, respectively. The regions of magnified images in the overview IHC sections were marked with red boxes.

Abbreviations: KO: cartilage-specific knockout of miR-128a; WT: wild-type; DMM: destabilization of the medial meniscus (∗∗∗p < 0.001) (male/female = 3/3 for WT, male/female = 3/3 for WT + DMM, male/female = 3/3 for KO, and male/female = 3/3 for KO + DMM).

Based upon the results mentioned above, the expression of NR1D2 decreased in both human OA samples and wild-type mice undergoing DMM procedures. However, the decrease in NR1D2 expression after DMM procedure was less apparent in miR-128a knockout mice. It is thus intriguing to speculate the correlation between miR-128a and NR1D2.

In order to determine whether NR1D2 is a possible target gene for miR-128a, we employed on-line software TargetScan (http://www.targetscan.org/vert_72) to identify candidate miRNAs that could potentially inhibit NR1D2 transcription in silico [24]. Based upon Watson-Crick base pairing based in silico prediction, NR1D2 mRNA 3’ UTR harbors binding sites for miR-128 [Fig. 8A]. We subsequently tried to determine whether activating NR1D2 signaling could affect chondrocyte function by treating chondrocyte cultures with miR-128a precursor (pre128) and/or NR1D2 agonist SR9009. Under RT-qPCR assay, we showed that NR1D2 agonist SR9009 mitigated miR-128a precursor-suppressed expression of aggrecan, collagen II, and sox9 [Fig. 8B ∼ D]. The extracellular matrix production by the chondrocyte micromass culture was significantly inhibited by miR-128a precursor, and the suppressed extracellular matrix by miR-128a could be rescued by NR1D2 agonist SR9009 [Fig. 8E].

Fig. 8.

The association between miR-128a, NR1D2, and cartilage anabolism. (A) The putative binding sites for miR-128 on the NR1D2 mRNA 3′-untranslated region (3′-UTR). (B ∼ D) The rescuing effects of NR1D2 agonist for miR-128a precursor suppressed expression of cartilage anabolic factors. (B) aggrecan; (C) collagen type II; and (D) SOX9 (SRY-box transcription factor 9).

Abbreviations: NC: normal control; pre-128a: miRNA-128a precursor (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001) (n = 4 for NC, n = 4 for pre-128a, and n = 4 for pre-128a + SR9009). (E) Alcain blue staining of the chondrocyte micromass culture under the treatment of miR-128a precursor and NR1D2 agonist SR9009.

Abbreviations: NC: normal control; pre128: miRNA-128a precursor (∗∗∗p < 0.001) (n = 8 for NC, n = 8 for SR9009, n = 8 for pre-128a, and n = 8 for pre-128a + SR9009).

Discussion

Increasing studies have demonstrated that miRNAs could be involved in the regulation of OA progression and in the regulation of circadian rhythm [3,6]. However, the network between OA, CCGs, and miRNAs has not been clearly defined. In our study, we showed that the transcription level of NR1D2 was lower in both human OA samples and samples from wild-type mice undergoing DMM procedure than in the non-OA counterparts. MiR-128a knockout mice were presented with less osteophyte volume and less disturbances in kinematic parameters 8 weeks after DMM procedure. Safranin-O staining showed that the histologic cartilage loss after DMM procedure was less severe in miR-128a knockout mice than in wild-type mice. IHC staining of NR1D2 showed that the decrease in the percentage of NR1D2 (+) chondrocytes after DMM procedure was less apparent in miR-128a knockout mice than in wild-type mice. We also showed that NR1D2 agonist could rescue the miR-128a precursor-suppressed expression of cartilage anabolic factors and extracellular matrix deposition. Based upon these findings, we proposed the signaling cascade between miR-128a and NR1D2 [Fig. 9].

Fig. 9.

Proposed signaling between miR-128a and NR1D2 and the progression of OA.

The impacts of disturbed circadian rhythm on OA progression are multifactorial and are not completely understood at present. Chronic circadian rhythm dysregulation could affect the inflammatory response, evidenced by the pro-inflammatory milieu of osteoarthritic joints after 24 h continuous light exposure [25]. Previous studies have shown that the expression of CCGs has an impact on cartilage metabolism (as shown in Table 1). These studies suggest a complex network of CCGs that work together to regulate cartilage metabolism. Additionally, the expression of these CCGs is subject to post-transcriptional control by miRNAs, further increasing the complexity of the network. Our study revealed decreased expression of NR1D2 in human OA samples and mice WT + DMM samples compared to human non-OA samples and mice WT samples, respectively. NR1D1 expression exhibited differential expression only in human OA and non-OA samples, but not in WT and WT + DMM mice samples. In contrast, the expression of BMAL-1, PER1, PER2, and PER3 showed no significant differences between human OA and non-OA samples, as well as mice WT and WT + DMM samples. Previous studies have also reported a decrease in NR1D1 expression in human OA samples compared to non-OA samples, suggesting that NR1D1 may also be a potential therapeutic target [26]. However, due to the insignificant effect of miR-128a knockout on the transcription level of NR1D1, the present study did not explore NR1D1 further. The transcription level of BMAL-1 was found to be different between OA and non-OA samples in a previous study, but our study did not observe such discrepancies [26]. Possible reasons for the observed differences in our study compared to a previous one could be the variation in the definition of non-OA samples. While the previous study utilized non-OA samples obtained from tissue banks with these non-OA samples processed within 24–72 h post-mortem, we collected cartilage tissues from regions without gross attrition during total knee arthroplasty to serve as non-OA samples. This variation in non-OA sample definition may also explain why we did not find a significant difference in BMAL-1 and PER2 expression between OA and non-OA samples, while the previous study reported decreased BMAL1 and increased PER2 expression in chondrocytes challenged with interleukin 1β, hydrogen peroxide, or basic calcium phosphate crystals compared to those without challenge [27].

Table 1.

The impact of CCGs expression on cartilage metabolism.

CCGs	Descriptions/Interpretations	References
Bmal1	NR1D1 and BMAL1 mRNA and protein levels were significantly reduced in OA compared to normal cartilage. Increased BMAL1 and decreased NR1D1 expression was noted after NR1D1 knockdown in cultured human chondrocytes. RNA sequencing from chondrocytes treated with NR1D1 or BMAL1 siRNA identified 330 and 68 different genes, and this predominantly affected TGF-β signaling.	[26]
Bmal1/Per2	Primary human normal cartilage-derived chondrocytes were challenged with IL-1β, HP or BCP crystals. Decreased BMAL1 and increased PER2 expression was noted after changes with any of 3 treatments. Levels of SOX9 were lower, whereas levels of ADAMTS5 and MMP13 were higher, in chondrocytes exposed to any of 3 treatments. PER2 siRNA partially abrogated the effects of each treatment on chondrocyte phenotype marker expression. In OA cartilage-derived chondrocytes, PER2 knockdown was associated with increased SOX9, reduced ADAMTS5 and reduced MMP13 expression. Further ablation of BMAL1 expression in OA chondrocytes resulted in a further reduction in SOX9 and increase in MMP13 expression. PER2 overexpression in chondrocyte cell line led to increased ADAMTS5 and MMP13 and decreased SOX9 expression. Localized inflammation, oxidative stress and BCP crystal deposition in OA joints may promote OA progression by inducing changes in the chondrocyte circadian clock.	[27]
Clock	The circadian clock could be reset by temperature signals, while the circadian period was temperature compensated. Fusion protein reporter PER2:luc bioluminescence showed that circadian oscillations were significantly milder in cartilage from aged mice. Time-series microarray analyses of the mouse tissue revealing that 615 genes (∼3.9% expressed genes) displayed a circadian pattern of expression, including genes involved in cartilage homeostasis and with potential importance in OA pathogenesis. Several clock genes were disrupted in the early stages of cartilage degeneration in the DMM OA model. These results reveal an autonomous circadian clock in chondrocytes, and circadian disruption (e.g., during aging) may compromise tissue homeostasis and increase OA susceptibility.	[28]
Cry2	In human OA cartilage, CRY2 staining and mRNA expression was significantly decreased. Cry2 was also suppressed in mice OA cartilage. Cry2 KO OA mice showed increased histologic severity in cartilage, subchondral bone and synovium. In OA chondrocytes, the CRY2 levels and the amplitude of circadian fluctuation were significantly lower. Pathway analysis that circadian rhythm and ECM remodeling were dysregulated in Cry2 KO mice. These results show the role of CRY2 in maintaining ECM homeostasis in cartilage, and targeting CRY2 has potential to normalize CCGs expression and reduce OA severity.	[29]

Abbreviations: NR1D1: nuclear receptor subfamily 1 group D member 1; BMAL1: basic helix-loop-helix ARNT like 1; TGF-β: transforming growth factor-beta; IL-1β: interleukin-1 beta; HP: hydrogen peroxide; BCP: basic calcium phosphate; PER2: period circadian clock 2; SOX9: SRY-box transcription factor 9; ADAMTS5: ADAM metallopeptidase with thrombospondin type 1 motif 5; MMP13: matrix metalloproteinase 13; RT-qPCR: reverse transcriptase-quantitative polymerase chain reaction; CRY2: cryptochrome circadian regulator 2; KO: knockout; ECM: extracellular matrix.

Our team has been interested in the role of miR-128a in the pathogenesis of OA. In our previous work, we showed that intra-articular injections with miR-128a antisense oligonucleotide stabilized chondrocyte autophagy and slowed anterior cruciate ligament-mediated articular tissue destruction [14]. In vitro, miR-128 signaling hindered Atg12 expression, LC3-II conversion, and autophagic puncta formation through targeting the 3′-UTR of Atg12. MiR-128a enhanced apoptosis, diminishing cartilage formation capacity. We highlighted that miR-128a induced Atg12 loss repressed chondrocyte autophagy to aggravate OA progression. In our current work, we aimed to expand our understanding of miR-128a′s role in regulating OA progression by examining its impact on the post-transcriptional regulation of CCG expression. Compared with the intra-articular injection of antisense nucleotide employed in the previous work, we employed cartilage-specific knockout model for more sustained suppression of miR-128a in our present work. We comprehensively checked the impact of miR-128a knockout on the expression of several major CCGs. Intriguingly, the expression of NR1D2 was most substantially suppressed by miR-128a knockout. It is unclear why miR-128a specifically affects NR1D2 expression instead of having a global impact on all CCGs. Our hypothesis is that the varying affinities between miR-128a and the 3′ UTR of different CCG mRNAs could potentially explain these findings.

NR1D2 also known as ERV-REBβ), is a ligand-dependent transcriptional repressor. However, NR1D2 could also activate Srebp-1c in skeletal muscle cells [30]. NR1D2 regulates circadian rhythm, metabolism and inflammatory response, has been found to be involved in various diseases, including cancer [31]. In a mouse OA model, the disruption of circadian rhythm was associated with hyperalgesia which could be reversed by the administration SR9009 (agonist for NR1D family), suggesting its potential as a novel analgesic [32]. In our study, we showed that the expression level of NR1D2 was lower in both human OA sample and wild-type mice undergoing DMM procedure. The decrease in the percentage of NR1D2 (+) chondrocytes after DMM procedure was less apparent in miR-128a knockout mice than in wild-type mice. We also tried to unravel the possible correlation between NR1D2 and miR-128a. We employed online software TargetScan (http://www.targetscan.org/vert_72) to identify candidate miRNAs that could potentially inhibit NR1D2 transcription in silico [24]. TargetScan identifies a perfect Watson-Crick base pairing complementary between 7 nucleotide long miRNA seed (from base 2 to 8 in the 5′ end of the miRNAs), and the annotated 3′-UTR (untranslated region) sequence and expends each seed match with additional base pairings to the miRNA. It further calculates the thermodynamic free energy of miRNA-target interaction. Based upon TargetScan-based in silico prediction, NR1D2 mRNA 3′ UTR harbors binding sites for miR-128. We also showed that NR1D2 agonist could restore the miR-128a precursor-suppressed expression of aggrecan, collagen type II and sox9. It is intriguing that the miR-128a/NR1D2 cascade affects the expression of multiple chondrogenic anabolic factors, suggesting that the manipulation of miR-128a/NR1D2 pathway could exert global cartilage anabolism. These findings presented researchers a promising novel target to treat OA.

There are strengths of our work. We offered novel piece of evidence to decipher the crosstalk between miRNAs, CCGs, and OA progression by employing the miR-128a knockout mice and DMM OA model. The impacts of miR-128a knockout and DMM on the expression of major CCGs were comprehensively investigated in our study. The are limitations to our study as well. Adhering to OARSI standards, it would be more appropriate to induce OA via DMM on mice that are ≥12 weeks old. We initially attempted to perform DMM in 12-week-old mice but found that the histologic severity was too severe 8 weeks after DMM in our mice. Therefore, we modified our protocol and performed DMM in 8-week-old mice, and we collected cartilage samples 8 weeks after DMM. The divergence of our protocol from the OARSI standards poses a risk to the dependability of our results. The impact of cartilage-specific miR-128a knockout on growth plate development was not demonstrated in the current paper and warrants further investigation. While we did observe a trend of differential cartilage thickness between WT and KO mice, we did not find any significant differences between KO and WT mice in terms of micro-CT related parameters, gait parameters, and quantitative histologic findings. Therefore, we cannot conclude that cartilage-specific miR-128a knockout enhances anabolic chondrogenesis based on these findings. One major limit of our work is the lack of sophisticated studies in detailing the genetic interplays between NR1D2 and miR-128a. Rigorous validation of the molecular interactions between NR1D2 and miR-128a will be warranted in the future studies and harbors great value in designing CCGs-based OA therapeutics.

Conclusions

Our study showed that cartilage-specific miR-128a knockout mice demonstrated less osteophyte formation, less disturbed gait kinematic parameters, and less suppression in NR1D2 expression after DMM procedures. NR1D2 enhancement could enhance the expression of cartilage anabolic factors. Our data provide new insights into how miR-128a signaling affects the crosstalk between cartilage metabolism and CCGs expression, highlighting the potential of miR-128a and CCGs targeting therapy to alleviate knee OA.

Role of the funding source

None.

Institutional Review Board statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Chang Gung Medical Foundation (IRB No.: 201802125B0, date of approval 2019/01/07). The animal study protocol was approved by the Institutional Review Board of Chang Gung Memorial Hospital (protocol code: 2018121501, date of approval 2019/04/17).

Informed consent statement

Informed consent was obtained from all subjects involved in the study.

Data availability statement

The data presented in this study are available on request from the corresponding author.

Conflicts of interest

None.