Regulation of microRNA‐221, ‐222, ‐21 and ‐27 in articular cartilage subjected to abnormal compressive forces

Key points microRNAs (miRs) are small non‐coding molecules that regulate post‐transcriptional target gene expression. miRs are involved in regulating cellular activities in response to mechanical loading in all physiological systems, although it is largely unknown whether this response differs with increasing magnitudes of load. miR‐221, miR‐222, miR‐21‐5p and miR‐27a‐5p were significantly increased in ex vivo cartilage explants subjected to increasing load magnitude and in in vivo joint cartilage exposed to abnormal loading. TIMP3 and CPEB3 are putative miR targets in chondrocytes Identification of mechanically regulated miRs that have potential to impact on tissue homeostasis provides a mechanism by which load‐induced tissue behaviour is regulated, in both health and pathology, in all physiological systems. Abstract MicroRNAs (miRs) are small non‐coding molecules that regulate post‐transcriptional target gene expression and are involved in mechano‐regulation of cellular activities in all physiological systems. It is unknown whether such epigenetic mechanisms are regulated in response to increasing magnitudes of load. The present study investigated mechano‐regulation of miRs in articular cartilage subjected to ‘physiological’ and ‘non‐physiological’ compressive loads in vitro as a model system and validated findings in an in vivo model of abnormal joint loading. Bovine full‐depth articular cartilage explants were loaded to 2.5 MPa (physiological) or 7 MPa (non‐physiological) (1 Hz, 15 min) and mechanically‐regulated miRs identified using next generation sequencing and verified using a quantitative PCR. Downstream targets were verified using miR‐specific mimics or inhibitors in conjunction with 3′‐UTR luciferase activity assays. A subset of miRs were mechanically‐regulated in ex vivo cartilage explants and in vivo joint cartilage. miR‐221, miR‐222, miR‐21‐5p and miR‐27a‐5p were increased and miR‐483 levels decreased with increasing load magnitude. Tissue inhibitor of metalloproteinase 3 (TIMP3) and cytoplasmic polyadenylation element binding protein 3 (CPEB3) were identified as putative downstream targets. Our data confirm miR‐221 and ‐222 mechano‐regulation and demonstrates novel mechano‐regulation of miR‐21‐5p and miR‐27a‐5p in ex vivo and in vivo cartilage loading models. TIMP3 and CPEB3 are putative miR targets in chondrocytes. Identification of specific miRs that are regulated by increasing load magnitude, as well as their potential to impact on tissue homeostasis, has direct relevance to other mechano‐sensitive physiological systems and provides a mechanism by which load‐induced tissue behaviour is regulated, in both health and pathology.


Introduction
Mechanical loading is essential with respect to regulating the functional capabilities of physiological systems including the musculoskeletal, cardiovascular and nervous system; this is achieved, at the cell and tissue level, by adapting to changes in mechanical load and/or metabolic stress applied. One of the major musculoskeletal tissues, articular cartilage, primarily functions to dissipate mechanical forces across the synovial joint surface and facilitates smooth, low-friction movement. The biomechanical integrity of articular cartilage is reliant on the biochemical composition of the extracellular matrix (ECM) , and maintenance of cartilage tissue homeostasis, effected by the chondrocytes, is similarly dependent on mechanical load (Buckwalter et al. 2005). Joint articular cartilage is predominantly exposed to dynamic compressive forces, although both tensile strain and shear stresses also result from everyday movement (Lee et al. 2005). Application of moderate, physiological mechanical loads is essential for maintaining cartilage homeostasis by promoting anabolic activities such as increased production of ECM molecules, whereas abnormal, non-physiological joint loading, as characterized by either overload or insufficient load, disrupts the homeostatic balance, favouring catabolism and cartilage degeneration, comprising the hallmark of osteoarthritis (OA) (Felson, 2013).
Mechano-regulation of cellular activities within physiological systems is known to occur through epigenetic mechanisms (e.g. RNA silencing). Primary contributors to RNA silencing are the microRNAs (miR), which are small (20-2 3bp), non-coding cytoplasmic RNAs that control the post-transcriptional regulation of one-third of all genes and are important in development, homeostasis and degeneration of tissues, including articular cartilage (Goldring & Marcu, 2012). Epigenetic studies have demonstrated that mechanical force has an impact on cellular responses through regulation of miR expression levels in tendon fibroblasts (Mendias et al. 2012), smooth muscle cells (Song et al. 2012), trabecular meshwork cells (Luna et al. 2011) and endothelial cells (Qin et al. 2010;Weber et al. 2010;Zhou et al. 2011). A small number of miRs were also identified as being mechanosensitive in chondrocytes (Dunn et al. 2009;Guan et al. 2011;Jin et al. 2014;Yang et al. 2016;Cheleschi et al. 2017). However, these studies were performed on isolated cells devoid of a substantial ECM, a feature known to be critical for cell-matrix mechano-communications (Guilak et al. 2006).
Therefore, using articular cartilage as a model system, the present study aimed to identify miRs that respond to 'physiological' and 'non-physiological' mechanical loads and to investigate the regulation of their potential downstream target genes.
to mechanical load. Primary chondrocytes were isolated from full depth cartilage utilizing the same tissue source as explants, and enzymatic digestion was performed (Al-Sabah et al. 2016). All cultures were maintained in 5% CO 2 and 20% O 2 at 37°C. Each experiment utilized tissue from between two and three animals, and repeat experiments utilized tissue from independent animals.

In vitro application of mechanical load to cartilage explants
Cartilage explants were subjected to either a 'physiological' (2.5 MPa, 1 Hz) or a 'non-physiological' load (7 MPa, 1 Hz) for 15 min using the ElectroForce 3200 (TA Instruments, New Castle, DE, USA) (Al-Sabah et al. 2016) and gene expression analysed at 2, 6 and 24 h post-load; unloaded explants served as controls. Explants were immediately snap frozen and remained in liquid nitrogen (<48 h) until RNA extraction. Loading regimes were selected based on articular cartilage literature demonstrating that ࣘ5 MPa is generally accepted as a 'physiological' load (Grodzinsky et al. 2000;Fehrenbacher et al. 2003), whereas peak loads >5 MPa are considered degradative (i.e. 'non-physiological') (Fehrenbacher et al. 2003); the frequency was set at 1 Hz, which has been demonstrated to resemble a human fast walking speed (Bader et al. 2011).

In vivo application of mechanical load
Twelve-week old male C57Bl6 mice (ß25 g; Envigo, Huntington, UK) were randomly assigned to either experimental or control groups and randomly allocated to MB1 cages (960 cm 2 ) in groups of five (12:12 h light/dark photocycles, with food and water available ad libitum). Animal husbandry and procedures were performed in compliance with the Animals (Scientific Procedures) Act 1986 [Home Office licence P287E87DF] according to Home Office and ARRIVE guidelines (Kilkenny et al. 2010). Mice were anaesthetized with isoflurane and custom-built cups used to hold the right ankle and knee in flexion with a 30 o offset prior to the application of a 0.5 N pre-load (ElectroForce13200; TA Instruments, Elstree, UK). A single 12 N load at a velocity of 1.4 mm s −1 was then applied resulting in anterior cruciate ligament (ACL) rupture as described previously ; mechanical loading was always conducted in the morning. Buprenorphine (0.05 mg kg −1 ) was administered S.C. to mice at the start of the experiment; animals were able to move freely and were monitored for welfare and lameness until termination of the experiment. Mice were culled by cervical dislocation at either day 1 or 7 post-load and the knee articular cartilage was dissected and processed for histology (toluidine blue staining) as described previously  or immediately snap frozen and remained in liquid nitrogen until RNA extraction. These early time points allowed assessment of mechanically regulated miRs in cartilage prior to overt degenerative changes and ECM loss. Nine animals were utilized for quantification of miR levels and the representative histology depicting the loading model phenotype is derived from experiments published in .

RNA extraction and reverse transcription for mRNA analysis
Total RNA was extracted from cartilage explants/ chondrocytes using 500 μL of Trizol reagent (Invitrogen, Paisley, UK) (Al-Sabah et al. 2016). RNA integrity was assessed (2100 Bioanalyzer and associated RNA 6000 Nano kit; Agilent Technologies, Wokingham, UK) and RNA integrity numbers >8.5 were observed. cDNA (total volume of 20 μL) was synthesized from 300 ng of total RNA using Superscript III reverse transcriptase in conjunction with 0.5 μg of random primers (Promega, Southampton, UK) in accordance with the manufacturer's instructions (Invitrogen).

RNA extraction and reverse transcription for miR analysis
Total RNA was extracted from cartilage explants/ chondrocytes as described above, except 1 mL of Trizol reagent was used. After ethanol precipitation, total RNA was purified using a mirVana miR Isolation Kit (Ambion, Paisley, UK) in accordance with the manufacturer's instructions. RNA integrity numbers of >8.0 were observed. cDNA of mature miRs was generated separately from total RNA (5 ng) using the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems, Paisley, UK) involving 50 U of MultiScribe Reverse Transcriptase and stem-looped reverse transcription primers, specific to individual miRs, from TaqMan MicroRNA Assays (Applied Biosystems, Paisley, UK) in accordance with the manufacturer's instructions.

miR next generation sequencing and bioinformatic analysis
Mechanically-regulated articular cartilage miRs were identified using next generation sequencing (NGS) using >3.5 μg of RNA per sample. Procedures were conducted in accordance with the manufacturers' instructions. Library preparation was conducted on 450 ng of total RNA using the NEB Next Small RNA Library Prep Set for Illumina (Multiplex Compatible: BioLabs, Hitchin, UK) and amplified cDNA was purified using a QIAquick PCR J Physiol 599.1 Purification Kit (Qiagen, Crawley, UK). miR libraries were selected by running purified cDNA samples on 8% (v/v) polyacrylamide gels and excising bands located at ß140 bp (Crowe et al. 2016). A Multiplex Compatible kit (NEB Next Small RNA Library Prep Set for Illumina) was used to elute and purify the miRs, and the concentration of miR libraries assessed prior to analysis on a HiSeq Sequencing System (The Genome Analysis Centre, Norwich, UK). miR deep sequencing data (raw FASTQ files) were run through FastQC and Cutadapt (Martin 2011), and trimmed FASTQ files were aligned against known bos taurus miR sequences from miRBase (http://www.mirbase.org). Quantification was determined by counting aligned reads against a reference, using a combination of RSamTools and ShortRead (Li et al. 2009) bioconductor packages. Differential expression was assessed using DESeq2 (Love et al. 2014). Global experimental variance was analysed using principal component analysis to assess for outlier samples and statistical significance from differential expression tests was determined by retaining miRs that had an adjusted P < 0.05.

Manipulation of miR expression levels in high-density chondrocyte cultures
Primary bovine chondrocytes were seeded onto six-well culture plates (VWR, Lutterworth, UK) at a density of 4 × 10 6 cells per well in antibiotic-free culture media and incubated at 37°C for 24 h prior to transfection. Chondrocytes were transfected for 48 h with 50 nM mirVana miR inhibitors (Applied Biosystems) or 50 nM miScript miR mimics (Qiagen) using DharmaFECT1 lipid reagent (Dharmacon, Cambridge, UK) in accordance with the manufacturer's instructions; mirVana miR Inhibitor Negative Control #1 (Applied Biosystems) and AllStars negative control small interfering RNA (siRNA) (Qiagen) were utilized as transfection controls (50 nM).

Quantification of miRNA and mRNA transcripts
Quantification of mRNA or miR in experimental samples was performed using a MxPro3000 QPCR system (Agilent Technologies, Stockport, UK) and measured using either reference gene primers (MWG-Biotech AG, Ebersberg, Germany) or bovine-specific TaqMan (Bustin et al. 2009). Cycling conditions were: 95°C-3 min (1 cycle), 95°C-15 s followed by 60°C-30 s (40 cycles) with an additional dissociation cycle of 95°C-1min, 60°C-30 s followed by 95°C-30 s (1 cycle) to confirm primer specificity with SYBR Green detection. Relative quantification was calculated using the 2 − CT method (Livak & Schmittgen, 2001), with unloaded controls as a reference group to quantify relative changes in transcript expression. Fold change was normalized to the geometric mean of 2-3 reference genes whose expression was identified as stable under the experimental condition using RefFinder software (https://www.heartcure.com.au/for-researchers/).

Luciferase activity assays
The 3 -UTR of mRNAs, containing the predicted binding site of target miRs, were cloned into pmirGLO dual-luciferase miRNA target expression vector (Promega, Southampton, UK) by In-Fusion (Takara Bio Europe, Saint-Germain-en-Laye, France) and construct sequences verified (for primer sequences, see Table 1). SW1353 chondrosarcoma cells (ß20 000 cells cm -2 ) were co-transfected with 50 nM miRNA mimics with the reporter plasmids (500 ng mL −1 ) (Barter et al. 2015); transfection of 50 nM AllStars negative control siRNA with the reporter plasmids was used as control. Following a 24 h transfection, cells were lysed and luciferase levels were determined using a Promega GloMax luminometer and the Dual-Luciferase reporter assay system (Promega).

Statistical analysis
Quantitative PCR (qPCR) data are presented as the mean ± 95% confidence intervals (CIs) after normalization to identified reference genes for explants (SDHA and YWHAZ), transfected cells (HPRT and YWHAZ) or in vivo model (U6, 18s and β-actin) and further normalized to untreated controls. Experiments were performed on explants (n = 6), transfected cells (n = 3) and in vivo studies (n = 9), with three independent repeats for explant and cell studies. Data were assessed for normality and differences in variances and transformed where required. One-way ANOVA and Fisher's post hoc test were performed to determine significance of mechanical load or manipulation of miR expression levels on gene expression, respectively; the results were considered statistically significant at P < 0.05 (Minitab, version 17; Minitab, LLC, State College, PA, USA).

miR target gene validation
Potential miR target genes identified by NGS were determined using Targetscan (http://www.targetscan.org) in conjunction with an assessment of their relevance to mechanical load or cartilage homeostasis as determined using the literature; putative target genes were examined by manipulation of expression levels using specific miR mimics or inhibitors (Fig. 3). Three putative miR-21-5p targets were selected: cytoplasmic polyadenylation element binding protein 3 (CPEB3), matrix metalloproteinase 13 (MMP13) and tissue inhibitor of metalloproteinase 3 (TIMP3). miR-221 and miR-222 seed sites are identical; hence, the selected putative target genes included: CPEB3, leukaemia inhibitory factor receptor (LIFR) and TIMP3. miR-27a target gene validation was not performed because a consistent reduction in miR-27a expression was not achieved using specific antagomirs. qPCR analysis confirmed that mimic-induced elevations in miR-221 levels resulted in a significant reduction in TIMP3 (P = 0.006) (Fig. 3A). Conversely, inhibition of miR-221 expression correlated with a significant increase in TIMP3 transcription (P = 0.003) (Fig. 3B). Similarly, a mimic-induced increase in miR-222 levels led to a significant reduction in TIMP3 (P = 0.006) (Fig. 3C).

Discussion
Physiological forces are critical for maintaining tissue homeostasis, and the involvement of epigenetic mechanisms such as mechano-regulation of miR expression occurs in many tissues, including articular cartilage (Dunn et al. 2009;Guan et al. 2011;Jin et al. 2014;Yang et al. 2015;Cheleschi et al. 2017). However, our understanding of miR involvement in  containing the predicted miR seed sites using a luciferase promoter assay SW1353 chondrosarcoma cells were co-transfected with reporter plasmids containing either (A) TIMP3 or (B) CPEB3 3 -UTRs and 50 nM miR-221, miR-222 or miR-21-5p mimics, or the negative control siRNA, for 24 h and luciferase levels were determined. Data are presented as the mean ± 95% CI (n = 3 wells) and are representative of three independent experiments. C, TIMP3 mRNA levels, as assessed using qPCR, in cartilage explants subjected to loads of 2.5 or 7 MPa (1 Hz, 15 min) and analysed 24 h post-cessation of load; unloaded explants served as controls. mRNA levels were normalized to the geometric mean of two reference genes (SDHA, YWHAZ) and further normalized relative to the unloaded control cDNAs. Data are presented as box plots depicting the mean ± 95% CI (n = 6 explants) and are representative of three independent experiments. Statistical analysis was performed using one-way ANOVA with Tukey's post hoc test.
response to different magnitudes of mechanical forces and, specifically, its impact on controlling mechanically induced tissue homeostasis is still in its infancy. The present study investigated the mechano-regulation of miRs in articular cartilage tissue explants subjected to 'physiological' and 'non-physiological' loads in vitro and validated regulated miRs in a murine in vivo model of abnormal joint loading. In addition, the study identified downstream miR targets to provide insight on mechanisms of mechanically mediated cartilage homeostasis. Importantly, the seed regions of the miRs of interest analysed in the present study are evolutionarily conserved across bovine, mouse and human species, indicating their potential physiological relevance.
Analysis of the miR-seq data illustrated that (i) a miR-mediated response to a 15 min loading episode was most noticeable at 24 h post-load and (ii) differentially regulated miRs were largely responsive to non-physiological compressive loads; the small number of miRs that were significantly regulated in response to physiological load probably reflects the loading regimen period. The miRs that were identified and validated to be most robustly altered by non-physiological load compared to unloaded controls and to physiological load were miR-221 and miR-222. This confirms the mechano-sensitive nature of miR-221 and miR-222, previously shown in cardiomyocytes after cardiac overload (El-Armouche et al. 2010), as well as in tendon fibroblasts (Mendias et al. 2012), engineered cartilage constructs in response to a catabolic loading regimen (Hecht et al. 2019) and anterior weight-bearing cartilage relative to the posterior non-weight bearing tissue in bovine stifle joints (Dunn et al. 2009).
Chondrogenic markers COL2A1 and SOX9 have been identified as putative gene targets for miR-221 and miR-222 (conserved seed site) that may influence cartilage homeostasis (Lolli et al. 2014); furthermore, miR-221 silencing strongly enhanced in vivo cartilage repair (Lolli et al. 2016). miR-221 inhibition also enhanced expression of chondrocyte-like phenotypic markers in intervertebral disc cells (Penolazzi et al. 2018). Therefore, miR-221 and miR-222 induction, observed in the cartilage explants in response to non-physiological load (Al Sabah A., Duance V. C., Blain E. J., unpublished observations), suggests an attempt to remodel the cartilage tissue to confer a more appropriate biomechanical response.
Analysis of downstream target genes identified robust regulation of TIMP3 only. However, although TIMP3 was clearly regulated via overexpression/inhibition studies in primary chondrocytes, this did not reflect observations in the SW1353 chondrosarcoma cell line for 3 -UTR activity using the luciferase assay or recapitulate events in tissue demonstrating that other, as yet unidentified, targets are regulated by miR-221 and miR-222 to elicit effects. These conflicting findings may be explained by the different experimental systems used in the present study, thus potentially masking the effects of other regulatory contributors with respect to the influence of miR-221 and miR-222 on Timp3 expression. Another possibility that might explain the simultaneous elevation of both the tested miRs and TIMP3 is a regulatory loop , such that elevated TIMP3 expression induces higher expression of these miRs to reduce Timp3 transcript levels in cells over time. Analysis at time points beyond 24 h post-load would provide insight as to whether potential regulatory loops exist.
Two other miRs robustly regulated by a magnitude-dependent load in our in vitro and in vivo loading models were miR-21-5p and miR-27a-5p. To the best of our knowledge, this is the first report of the mechano-regulation of these miRs in articular cartilage. However, miR-21 mechano-regulation occurs in other cell types; tensile strain induced miR-21 expression in human aortic smooth muscle cells (Song et al. 2012), and both laminar (Weber et al. 2010) and oscillatory shear stress (Zhou et al. 2011) elevated miR-21 levels in human umbilical vein endothelial cells. By contrast, pulsatile shear stress inhibited miR-21 expression in these endothelial cells (Zhou et al. 2011), revealing the mechano-sensitive nature of these molecules. In the present study, both TIMP3 and CPEB3 were identified as downstream targets of miR-21-5p; however, as noted previously, TIMP3 is not negatively correlated with miR-21-5p levels in our model systems. Furthermore, CPEB3 levels were not significantly regulated in the present study, indicating that, although these genes are direct targets of miR-21-5p in primary chondrocytes, they are not directly regulated in our models. As a result of the complexities of such signalling mechanisms in the tissue, further experiments are clearly required to determine the interplay of these miRs and their mechanistic activities in cartilage homeostasis, which both remain beyond the scope of the present study. miR-27a-5p was robustly regulated by mechanical load both in vitro and in vivo. Mechano-regulation of miR-27a in articular cartilage is a novel finding and corroborates studies demonstrating up-regulation of both miR-27a and miR-27b in endothelial cells subjected to laminar flow (Urbich et al. 2012) and endothelial cells exposed to cyclic tensile strain . Downstream targets of miR-27-5p, which are known to be regulated in in situ cartilage explants in response to non-physiological load Al-Sabah et al. (unpublished data), include WNT signalling molecules such as DKK2 (Tao et al. 2015;Wu et al. 2019) and sFRP1 (Wu et al. 2019). Future studies will explore the relationship between mechano-sensitive miR-27-5p and downstream regulation of WNT signalling components in cartilage homeostasis.
A reduction in miR-483 levels was observed in response to non-physiological load and is the first report of its mechano-sensitivity in articular cartilage. Its potential role in cartilage homeostasis is not well defined, with conflicting evidence suggesting anabolic (Yang et al. 2015) and catabolic outcomes (Xu et al. 2017;Wang et al. 2019); hence, its observed reduction in response to abnormal load may reflect an attempt at tissue remodelling.
A limitation of the present study is use of immature articular cartilage removed from underlying subchondral bone, which could influence mechano-biological outcomes. However, to mitigate this limitation, we validated identified miRs in an in vivo model of abnormal joint loading to confirm their mechano-regulation; interestingly, many of the miRs regulated by load in our in vitro and in vivo models have also been reported to be differentially expressed in OA (Tardif et al. 2009;Zhang et al. 2014;Song et al. 2015;Wang et al. 2019), lending weight to their relevance in cartilage homeostasis.
In conclusion, the loading magnitude-dependent regulation of specific miRs identified in the present study, as well as their potential to impact on tissue homeostasis, has direct relevance to other physiological systems that are mechano-sensitive. Furthermore, it provides a pivotal mechanism by which load-induced tissue behaviours are regulated, in both health and pathology, and is critical to understand with respect to successful tissue engineering strategies in physiological systems.