The sequences of the identified conserved and novel cabbage miRNAs (11) were used to predict their potential human target genes. The Homo sapiens 3′UTR, 5′UTR and CDS sequences, that were downloaded from the UCSC Genome Bioinformatics Site (http://genome.ucsc.edu/) and NCBI CCDS Database (www.ncbi.nlm.nih.gov/CCDS/CcdsBrowse.cgi), respectively, served as reference target genes dataset. Bioinformatics targets prediction was performed using miRanda method (http://cbio.mskcc.org/miRNA2003/miranda.html), the search procedure of which takes into accounts sequence complementarity, interspecies conservation and thermodynamic stability of the miRNA:mRNA duplex. The prediction parameters were as follows: 1) G:U base pairing was permitted but scored lower (score +2) than canonical base pairs (score +5), 2) the alignments with gaps and non-canonical base pairs in the “seed” regions (2–8 nt at the 5′ end of the molecule) were discarded, and 3) alignments with scores over 130 and minimum free energy (MFE) of the structure less than −17 kcal/mol were selected. Generated list of putative target genes was sorted by the highest alignment score and lowest MFE of the structure. To designate potential processes involving the predicted mRNA sequences and to asses a probable anti-inflammatory effect of cabbage miRNAs on human organism, the selected targets were mapped on the UniProt database (www.uniprot.org) and annotated using the Blast2GO (http://www.blast2go.com/b2ghome) software. The obtained dataset was analyzed and the most interesting human target gene for cabbage bol-miR172 was selected.

The miR172a and miR172a* sequences were synthesized on the basis of bol-miR172a from Brassica oleracea var. capitata (Lukasik et al., 2013): 5′-AGA​AUC​UUG​AUG​AUG​CUG​CAU-3′.

The mu_miR172a and mu_miR172a* sequences were designed by changing 3 nt (underlined) inside the region corresponding to the mouse FAN mRNA sequence potential binding site to preserve complementarity: 5′-AGA​AUC​UUA​AGA​AUG​CUG​CAU-3′.

The miR172 mimics were chemically synthesized in the form of single-stranded or miR172a/miR172a* and mu_miR172a/mu_miR172a* duplexes with both 5′ RNA strands phosphorylated and 3′ ends protected by 2′-O-methylation, and HPLC purified (FutureSynthesis, Poland).

Human foreskin fibroblasts (K21; NCBI GEO: GSM1227449) were cultured in MEM/EBSS medium (HyClone) supplemented with nonessential amino acids (Gibco) and 10% inactivated FBS (Lonza) at 37°C in a 5% CO₂ atmosphere.

Mouse embryonic fibroblasts (WT MEFs; ATCC Manassas, United States) were cultured in DMEM (Gibco) supplemented with 10% FBS (Gibco), 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco) at 37°C in a 5% CO₂ atmosphere.

Details of the transfection procedures are given in the Supplementary Material.

Fibroblasts K21 were plated on 6-well plates (2.8 × 10⁵ cells/well) and the next day transfected with the miR172a-lipofectamine RNAiMAX (Invitrogen) complex (final concentration of miRNA when added to cells was 60 pM, 1 nM or 10 nM). MISSION^® microRNA Mimics HMC0003 (Sigma) and Hs_NSMAF_3 (QIAGEN) were used as a negative and positive control, respectively at a final concentration of 10 nM. The cells were incubated for 48 or 72 h. Four hours before the end of the experiment, LPS (final concentration 1 μg/ml) was added to each well. Then, the cell pellet was suspended in 400 µl of PBS and stored at −20°C until analysis.

MEFs were plated on 24-well plates (1.9 × 10⁴ cells/well) and the next day, 1 nM, 10 nM or 100 nM mu_miR172a/mu_miR172a* duplex was transfected into MEFs (two wells each) using ScreenFect A-plus (ScreenFect GmbH) according to the manufacturer’s instructions. The next day, the transfection procedure was repeated, and the cells were incubated for the next 24 h. As a negative control, nonspecific dsRNA (siGENOME Nontargeting siRNA #3, Dharmacon) at a concentration of 10 nM was used. Four hours before the end of the experiment, TNF-α (1 ng/ml) was added to each well. Then, cell pellets from two wells with the same transfection conditions were suspended in 400 μl of PBS and stored at −20°C until analysis.

Transfection efficiency was determined microscopically for each experiment by using 10 nM BLOCK-iT™ Alexa Fluor^® Red Fluorescent Control (Invitrogen).

The concentration of FAN protein was determined using NSMAF Human ELISA Kit (Cloud-Clone Corp.) for human fibroblasts or ELISA kits (EIAab) for mouse fibroblasts, respectively, according to the manufacturer’s protocols. Absorbance measurements were performed on a BioTek SynergyHT plate reader with the wavelength set at 450 nm. The concentration of FAN in the tested samples was calculated based on the standard curve and expressed in ng/mL.

Twenty DBA/1 male mice were used in the experiments, conducted according to Local Ethical Committee consent (Act No. WAW2/083/2019). Seven-week-old mice (Janvier Labs) after 1 week of adaptation, were randomly divided into four experimental groups. Experimental animals were subjected to both CIA induction and mu_miR172a administration (CIA + miRNA), n = 8. Control animals were 1) not treated (NT), n = 4; 2) subjected to mu_miR172a administration (miRNA), n = 4; or 3) subjected to CIA induction (CIA), n = 4.

The Experimental Design is Presented in Figure 1.

In experimental animals, CIA was induced according to protocols (Brand et al., 2007; Bevaart et al., 2010b). On Day 0, animals were subjected to anesthesia (ketamine, 100 mg/kg, xylazine, 10 mg/kg, intraperitoneally). Next, mice were injected subcutaneously into the tail, 1 cm below the tail base, with 100 µl of immunization mixture (1 mg/ml bovine collagen II, Chondrex, in complete Freund’s adjuvant, 2 mg/ml, DIFCO).

On day 20 (3rd week), animals were anaesthetized as previously and injected subcutaneously, 0.5 cm below the previous injection, with100 µl of booster immunization mixture (1 mg/ml bovine collagen II in incomplete Freund’s adjuvant, Santa Cruz).

The immunization mixtures were prepared by intense mixing between two syringes connected with luer-luer connections of collagen II solution with complete or incomplete Freund’s adjuvant (1:1) until the consistency of thick cream was obtained.

Two days before the booster injection, experimental animals received a subcutaneous injection (into the neck area) of mu_miR172a (0.2 mg/ml in 0.9% NaCl). Injection was repeated 2 days after the booster injection and repeated once a week. Animals received a total of six doses of mu_miR172a.

Beginning a week preceding booster injection (2nd week), once a week, paw thickness was measured. Joints of both left and right and front and hind paws were measured: metacarpus (height and width), metatarsus (height and width), and ankle joint (height and width). Means of 1) width and 2) height of both the left and right (L + R) metacarpus, metatarsus and ankle joint of animals from each experimental group in the following weeks were calculated. Finally, the ratio of width and height of both the left and right (L + R HxW) metacarpus, metatarsus and ankle joint of animals from each experimental group in the following weeks was calculated.

Along with paw thickness measurements, the body weights of the mice were obtained. Means of the body mass of animals from each experimental group in the following weeks were quantified.

Measurements of paw thickness and body mass of mice, which are the indicators of the disease development and treatment efficiency, were subjected to statistical analysis. Ratios of width and height of both the left and right (L + R HxW) metacarpus, metatarsus and ankle joint of animals from each experimental group in the following weeks (see Materials and Methods section: Paw thickness measurement) were counted for comparison.

The mean and the standard deviation (SD) for variables (ratios and body mass) were reported in the following weeks of the experiment. The variables were compared with the two-way ANOVA test or mixed models (in case of missing values). The significance level was set at 0.05. The statistical significance is marked in the figure as follows: *p < 0.05; **<0.01; ***p < 0.001.

To determine the influence of plant miR172a on the regulation of mammalian FAN protein expression, we examined the effect of synthetic miR172 mimics on the level of FAN protein in selected human and mouse cell lines upon inflammation-inducing factor treatment.

We first examined the level of FAN protein in human foreskin fibroblast K21 cells after stimulation by LPS or TNF-a treatment. As shown in Supplementary Figure S1, the expression of FAN was significantly higher upon LPS treatment. Thus, subsequent experiments were conducted using LPS as an inducing agent. It was shown that the effect of miR172a addition was more pronounced in cells incubated for prolonged periods after transfection (24 vs. 72 h; Supplementary Figure S2). Additionally, single-stranded miR172a was almost ineffective in comparison with the same double-stranded molecule (Supplementary Figure S3).

Figure 2A shows summarized data of FAN levels obtained after 48 h of incubation of K21 cells stimulated with LPS. A 10 nM concentration of duplex miR172a caused the greatest reduction in FAN levels when compared with negative or untreated controls. Notably, a concentration as low as 1 nM ds miR172a resulted in a decrease in FAN levels below the level acquired with positive control siRNA.

In further analyses, we tested the effect of the miR172 mimic, the customized mu_miR172a sequence, on mouse cells.

MEFs were transfected with different concentrations of mu_miR172a/mu_miR172a* duplexes (twice every 24 h) and then stimulated with TNF-α. As shown in Figure 2B, the presence of the mu_miR172a duplex markedly decreased the level of FAN protein in TNF-a-treated MEFs compared to cells not transfected with any miRNA or transfected with nonspecific dsRNA. The concentration of 10 nM mu_ miR172a was the most efficient among all tested and therefore became the basis for calculating the miRNA mimic dose administered to the experimental animals.

It should be mentioned however, that the above described experiments were performed repeatedly, and a similar dependency between the tested miR172 mimics and controls was observed. As experiments differed in the level of transfection efficiency and level of cell viability, we decided to present data from one representative experiment instead of combining data from a few experiments into one graph.

Having found that plant-derived miR172a mimics modulate the level of FAN protein in human and mouse cells, we undertook preliminary trials in an animal model of rheumatoid arthritis.

To evaluate the anti-inflammatory effect of mu_miR172a in animals, we administered mu_miR172a subcutaneously to mice with an evoked model of rheumatoid arthritis (CIA).

Beginning at the 2nd week after inducing CIA, body mass and paw thickness in mice were measured. Figure 3 presents the comparison of the metacarpus and metatarsus of control animals and metacarpal and metatarsal oedema observed in experimental (treated and untreated) mice.

The metacarpus (A), metatarsus (B) and ankle joint (C) size and mean body mass (D) of control and experimental (treated and untreated) mice in following weeks were presented on the Figure 4.

There were no differences in the size (ratios of height and width) of the metacarpus, metatarsus or ankle joint (Figures 4A–C respectively) between healthy not treated (NT) animals and healthy animals treated with mu_miR172a (miRNA) during the experiment.

Beginning at week 3, the size of the metacarpus (Figure 4A) of CIA mice was significantly increased compared to that of healthy mice (NT, miRNA). In the following weeks, except for the 6th week, the size of the metacarpus of CIA mice was significantly increased compared to that of all other experimental groups (NT, miRNA, CIA + miRNA). There were no significant differences in the size of the metacarpus of CIA mice treated with mu_miR172a (CIA + miRNA) compared to healthy animals (NT, miRNA), except for the 6th week, where the size was increased significantly in comparison to healthy, mu_miR172-treated mice (miRNA).

At the 4th week, there was a significant increase in the size of the metatarsus (Figure 4B) of CIA mice, both not treated (CIA) and mu_miRNAa treated (CIA + miRNA), compared to healthy animals (NT, miRNA). This was observed for 2 weeks for CIA mice (CIA) and until the end of the experiment for CIA mice treated with mu_miR172 (CIA + miRNA). At the 4^th week, the size of the metatarsus of CIA mice (CIA) was significantly increased compared to that of CIA mice treated with mu_miR172a (CIA + miRNA).

Beginning at the 4th week, there was a significant increase in the size of the ankle joint (Figure 4C) of CIA mice (CIA) compared to all other experimental groups (NT, miRNA, CIA + miRNA). Beginning with a 5th week increase in the size of the ankle joint in CIA mu_miR172a-treated mice (CIA + miRNA) compared to healthy animals (NT, miRNA), no significant difference between nontreated (CIA) and mu_miR172a-treated (CIA + miRNA) mice was observed until the end of the experiment.

During the course of the experiment, the body mass (Figure 4D) of the healthy (NT, miRNA-treated) animals increased slightly similarly. The CIA animals with administered mu_miR172a (CIA + miRNA) showed no changes in body mass. A significant drop in body mass was observed in CIA animals (CIA) after the booster injection in the 3rd week compared to other experimental groups, remaining lower until the end of the experiment.