Chondrocytes were isolated from male 100 to 125 g Sprague Dawley rats as has been previously detailed by Boyan et al. [3, 21]. Animal procedures connected with this work were approved by the Institutional Animal Care and Use Committee at Virginia Commonwealth University. In brief, rats were killed by CO2 asphyxiation followed with cervical dislocation. Sharp dissection was used to remove the rib cages and trim excess tissue. Ribs were placed in Dulbecco’s Modified Eagle’s Medium (DMEM) (Life Technologies, Carlsbad, CA) containing 1 g/L glucose, 150 U/mL penicillin and 150 μg/mL streptomycin and kept on ice. Under a dissection microscope all tissue was cut from around the ribs. The bone, GC cartilage, and RC cartilage were clearly visible under the microscope and the cartilage was carefully cut into slices using a scalpel. One or two transition slices between the RC cartilage and GC cartilage and between the GC cartilage and bone were discarded and the remaining GC slices were incubated (37˚C and 5% CO2) overnight in DMEM containing 1 g/L glucose, 50 U/mL penicillin, and 50 µg/mL streptomycin, plus 10% fetal bovine serum (FBS). The following day the slices were washed twice with Hank’s Balanced Salt Solution containing 50 U/mL penicillin and 50 μg/mL streptomycin then incubated in 0.25% trypsin–EDTA (Gibco, Gaithersburg, MD) for 1 h, washed once as before and incubated (37˚C and 5% CO2) in 0.2% collagenase type II (Worthington Biochemical, Lakewood, NJ) on a shaker for 3 h to digest the ECM. The resulting cell suspension was passed through a 40 μm nylon mesh strainer; FBS was added to 10% and the cells were pelleted by centrifugation (500 × g, for 10 min) followed by resuspension with DMEM Full Media (DMEM FM; DMEM with 10% FBS, 1 g/L glucose, 50 U/mL penicillin, 50 µg/mL streptomycin, and 50 µg/mL ascorbic acid). Cells were counted and then plated at a density of 20,000 cells/cm2. Cells were incubated (37°C and 5% CO2) and media were changed 24 h after plating and then every 48 h. Fourth passage cells were used for all experiments; GC chondrocytes at this passage retain their GC phenotype and provide sufficient cells for MV isolation from the ECM.

We selected five microRNA that were highly exported into MVs from chondrocytes based on previous experiments [14, 19]. These microRNA exhibited fold changes ranging from 2.19 to 2,352.53 when comparing the GC microRNA populations between the cell isolation and the MVs that they produced (Fig. 1A). Three of the microRNAs were upregulated in the MVs following treatment of the chondrocytes with 1α,25(OH)2D3 and two of the microRNAs were highly enriched in the MV without treatment. Triple stranded locked nucleic acid (LNA) mimics of these microRNA biotinylated at the 3′ end were purchased from Qiagen (Hilden, Germany). In addition, mirVana™ miRNA mimics were purchased from (Thermo Fisher, Waltham, MA).

MicroRNA expression in matrix vesicles (MV) and cells and mRNA clustering after LNA based pulldown. A Fold changes in microRNA expression between MV and Cells. B 2D principal component analysis (PCA) of all 13,664 sequenced genes with PC1 on x-axis and PC2 on y-axis. C 2D PCA of the 502 differentially expressed genes with PC1 on x-axis and PC2 on y-axis. D 3D PCA of the 502 differentially expressed genes adding PC3 on the z-axis. Two different angles are displayed demonstrating separation between the six groups

In order to determine the optimal transfection concentration, we transfected fourth passage GC cells at 70% confluence with the LNA scrambled negative control (NC) (14.5 nM) or with different amounts of the LNA microRNA mimics 22-3p and 122-5p and measured DNA content, as described below. The scrambled NC and microRNA mimics 22-3p and 122-5p solutions were diluted 1.5X four times before combining with lipofectamine solution (final concentrations of 32.6, 21.8, 14.5, 9.7, and 6.5 nM). Cells were treated with transfection solution for 24 h before changing media to DMEM FM for another 48 h of incubation before DNA isolation and quantification. We also transfected with mirVana™ NC and microRNA mimics 22-3p and 122-5p at 14.5 nM, which we had determined in a previous study (purchased from Thermo Fisher, Waltham, MA) [14]. This was necessary as we were switching systems to a triple stranded LNA mimic and wanted to validate similar phenotypic responses in our cells that correspond to previous work.

Based on these results, fourth passage GC chondrocytes were grown to 70% confluence, media were aspirated and replaced with DMEM 1X with 10% FBS. LNA mimics were diluted 100-fold to 666.7 nM in DMEM and combined 1:1 with a solution of lipofectamine RNAiMAX transfection reagent (Invitrogen, Carlsbad, CA) that had been diluted 26-fold in DMEM. This solution of mimics and lipofectamine sat at room temperature for 20 min. Cells were treated with transfection solution for final LNA concentration of 32.6 nM and incubated in transfection media for 48 h before RISC pulldown and RNA isolation.

DNA quantity was assayed and compared with previous results to determine optimal transfection concentrations of the new type of microRNA mimics. The media were aspirated and the cell layers washed twice with 1X PBS. 100 µL of 0.05% Triton- × 100 in H2O was added per well and the plate moved to −80 °C for storage. Samples were thawed on ice and sonicated (40 amps, 10 s per well). The QuantiFluor dsDNA system (Promega, Madison, WI) was used for quantification following standard protocol with samples diluted 1:10 and read on a plate reader (Synergy H1 Hybrid Reader, BioTek, Winooski, VT) with excitation of 485 nm and emission of 538 nm.

RISC and RNA isolation were performed using a protocol based on Dash et al. [22] 30 µL per sample of streptavidin coated magnetic beads (Pierce, Waltham, MA) were washed the day before cell harvest (3 washes with 100 µL of 10 mM Tris–Cl pH 7.5, 0.5 mM EDTA, 1 M NaCl solution and 3 washes with 100 µL solution of 0.1 M NaOH, 0.05 M NaCl), resuspended (100 µL solution of 0.5 M NaCl), and blocked (200 µL of 1 μg/μL BSA, 2 μg/μL yeast tRNA solution) overnight in 4 ˚C. Cells were removed from flasks with trypsin, pelleted (1500 × g for 5 min at 4  C), washed with sterile 1X Dulbecco’s phosphate buffered saline (DPBS) (Cytiva, Marlborough, MA), pelleted again as before, and resuspended in 600 µL lysis buffer [150 mM NaCl, 25 mM Tris–HCl pH 7.5, 5 mM DL-dithiothreitol (DTT) (Sigma-Aldrich, St. Louis, MO, USA), 0.5% octylphenoxy poly(ethyleneoxy)ethanol (IGEPAL) (Sigma-Aldrich, St. Louis, MO, USA), 60 U/mL Superase, 1 × protease inhibitor cocktail] before being rapidly frozen in −80 °C and then thawed on ice. Cellular debris was pelleted (16,000 × g for 5 min at 4 °C) and the supernatant combined with ¼ volume of 5 M NaCl to generate cell lysate solution. Beads from previous day were washed three times with 150 µL of pulldown wash buffer (10 mM KCl, 1.5 mM MgCl2, 10 mM Tris–HCl pH 7.5, 5 mM DTT, 1 M NaCl, 0.5% IGEPAL, 60 U/mL Superase, 1 × protease inhibitor cocktail) and then resuspended in 300 µL of pulldown wash buffer. 300 µL of cell lysate solution was incubated with 300 µL streptavidin coated beads for one hour at room temperature. Beads were washed three times with 300 µL pulldown wash buffer and finally resuspended in 100 µL of nuclease free water on ice. 700 µL qiazol (Qiagen) was added to each tube before transferring to −80 °C. RNA was precipitated following miRNeasy micro kit (Qiagen). RNA was eluted in 30 μL of nuclease free water per sample.

RNA isolations were quantified with a RNA 6000 pico chip on a BioAnalyzer (Agilent, Santa Clara, CA) and library prepared according to specifications of the PCR-cDNA Barcoding kit (SQK-PCB109, Oxford Nanopore, Oxford, UK) for 12 samples and run on a Spot On Flow Cell Mk 1 R9 (Oxford Nanopore) for 72 h. High precision nucleic acid basecalling was carried out on Nanopore’s MinIT running MinKNOW (21.05.24) and guppy (5.0.16) to determine the sequence.

Fastq files that were considered to have passed basecalling were transferred to VCU’s high performance research computing core facility where reads were aligned using minimap2 (2.21-r1071) to Ensemble’s Rattus norvegicus transcriptome (Rnor_6.0) and quantified with salmon (v1.5.2) [23, 24]. The count data were then analyzed in R (4.1.0), differential expression determined using DESeq2 (1.32.0), PCA plots built with pca3d (0.10.2) and PCAtools (2.4.0), Venn diagrams made with VennDiagram (1.6.20), heatmap with pheatmap (1.0.12), and volcano plots and histograms with ggplot2 (3.3.5) [25,26,27,28,29,30,31]. UTRdb was used to download the 3′ untranslated regions (UTRs) of the differentially expressed genes [32]. Specific microRNA sequences were downloaded from mirBase.org [33]. The RNAhybrid tool was used to determine the microRNA to mRNA 3′UTR minimum binding energies [34]. Pathway analysis was carried out using the PANTHER classification system (16.0) webtool [35].

DNA quantity is presented as mean ± standard error of the mean for six samples for each group. An ANOVA with Tukey HSD post-hoc test was used to examine differences between the groups. Significance was determined by a p > 0.05.