This study included 40 BC female patients that were subjected to conservative mastectomy or lumpectomy surgery in Kasr El-Aini and National Cancer Institute hospitals. Tumors and their noncancerous adjacent tissues were resected. Tissues were immediately frozen in liquid nitrogen and stored at –80°C. Pathology experts confirmed the pathological identity of all samples. The mean age of recruited patients was 47.5 years (Range 34–68) (Table 1, Supplementary Table S1). All patients agreed and signed an informed consent. The study was approved from the ethical committee of the German University in Cairo, Cairo University and National Cancer Institute hospital and was performed in compliance with the ethical standards of the declaration of Helsinki.

MDA-MB-231 TNBC cell line was obtained from VACSERA, Cairo, Egypt. Cells were cultured in DMEM (Dulbecco’s Modified Eagle Medium) (Serox GmbH, Germany) containing fetal bovine serum (FBS) (10%) (Lonza, Germany) and penicillin/streptomycin (1%) (Lonza, Germany) at 37°C under 5% CO₂ (Youness et al., 2019). Cells were grown in monolayer and passaged routinely 1–2 times a week as previously described (Youness et al., 2021).

Aerial parts of Rosmarinus officinalis L. (Rosemary) were collected from Al-Orman Garden, Egypt in May 2019. It was authenticated by Dr. M. Gebali, Taxonomy specialist at National Research Centre, Giza, Egypt. A voucher specimen of the authenticated plant (voucher No. 125) was deposited at herbarium of Pharmaceutical Biology department, Faculty of Pharmacy and Biotechnology, German University in Cairo, Cairo, Egypt.

The aerial parts of rosemary (5 kg) were powdered and exhaustively extracted using 85% ethanol (10L) by cold maceration for 3 days. This process was repeated for three times. The total extract was evaporated under reduced pressure using rotary evaporator (BUCHI, Rotavapor, R-210; Switzerland) at temperature 50°C–60°C to yield brownish semi-solid residue (210 g). This residue was suspended in distilled water, and subjected to liquid-liquid fractionation. Then fractionated using solvents with increasing polarity. The obtained residues were washed with acetone and lyophilized to obtain crude contents yielding 39 g and 45 g of dichloromethane and ethyl acetate fractions, respectively.

The analysis of the ethyl acetate fraction was done using mass spectrometric tool (UPLC-PDA-ESI- MS) through Waters ACQUITY Xevo TQD system, which consisted of an ACQUITY UPLC H-Class system and Xevo TQD triple-quadrupole tandem mass spectrometer with an electrospray ionization (ESI) interface (Waters Corp., Milford, MA, USA). Acquity BEH C18 column of 1.7 μm particle size with dimensions 100 mm × 2.1 was used for the separation (Waters, Ireland). The mobile phase consisted of 0.1% formic acid in water to represent the (A) and 0.1% formic acid in acetonitrile to represent (B) using a gradient elution as follows: 0–4 min, 15% B; 4–8 min, 20% B; 8–30 min, 55% B; 30–35 min, 90% B; and 35–40 min, 15% B. The used flow rate was 200 μl/min and injection volume was 10 μl. The eluted compounds were detected at mass ranges from 100 to 1,000 m/z. MS scan was carried out at the following conditions: capillary voltage, 3.5 kV; detection at cone voltages, (20 V–95 V); radio frequency (RF) lens voltage, 2.5 V; source temperature, 150°C; and desolvation gas temperature, 500°C. Nitrogen was used as desolvation and cone gas at a flow rate of 1,000 and 20 L/h, respectively. System operation and data acquisition were controlled using Mass Lynx 4.1 software (Waters).

The EA fraction (45 g) was applied over an open column (64 × 5.5 cm) ID packed with polyamide 350 g as stationary phase. Elution was done using methanol CH₃OH:H₂O gradient to ensure a complete elution process. The fraction exhibited a major blue spot on thin layer chromatography (TLC) using CH₃OH:H₂O: nitrous acid 6:3:1 under short wavelength Ultraviolet (UV) (254 nm). It then turned into yellow upon exposure to NH3 vapor. Further column was built up for purification of the major spot, then subjected to HPLC/PDA/ESI/MS analysis to identify its fragmentation pattern. Structure was elucidated and confirmed using Bruker 400 MHz Avance III HD, Switzerland. 1H-NMR (400 MHz) and 13C-NMR (100 MHz) spectra were recorded in deuterated DMSO using TMS as an internal standard and chemical shift values were expressed in δ ppm.

Three different target prediction algorithms TargetScan (http://www.targetscan.org/vert_72/), miRwalk (http://mirwalk.umm.uni-heidelberg.de), and starbase (http://starbase.sysu.edu.cn) were used to search for miRNAs that dually target IL-10 and TNF-α genes.

miR-17-5p mimics, MALAT1 siRNAs and H19 siRNAs were purchased from Qiagen, Germany. Cells were trypsinised, counted and seeded into six-well plates the day before transfection to ensure 80-90% cell confluence on the day of transfection (Youness et al., 2016a). Transfection into cells was performed using HiPerfect^® Transfection Reagent (Qiagen, Germany) in accordance with the manufacturer’s advised procedure as previously described (Youness et al., 2016a; Youness et al., 2016b; Youness et al., 2018). All Experiments were run in triplicates and repeated at least three times.

Biazol reagent (Hangzhou bioer technology Co., Ltd., China) was utilized for total RNA extraction from BC patients’ tissues and the TNBC cell lines. The resultant yield of RNA was assessed using a spectrophotometer.

Reverse transcription of TNF-α, IL-10, MALAT1, H19, β-actin and 18s rRNA mRNAs into complementary DNA (cDNA) was performed using a high-capacity cDNA Reverse-Transcription Kit (ABI, California, USA) according to the company's instruction. TaqMan_MicroRNA Reverse-Transcription Kit (ABI) was used to reverse-transcribe extracted miRNAs and specific primers for hsa-miR-17-5p and RNU6B into single-stranded cDNA. Relative expression of TNF-α, IL-10, MALAT1 and H19 was calculated after normalization to β-actin and 18s rRNA as reference genes. miR-17-5p and RNU6B (for normalization) cDNA were quantified using a TaqMan Real-Time qPCR on StepOne™ Systems (ABI, California, USA) (Youness et al., 2019). Relative expression was calculated using the 2^–ΔΔCT method. All PCR reactions were run in triplicates and repeated at least three times.

Cytokine levels were measured by ELISA technique. After transfection of MDA-MB-231 cell lines with miR-17-5p for 48 h, supernatants were collected and stored at –20°C until assayed for cytokines by ELISA (Thermo scientific, USA) as per manufacturer’s protocol.

Cellular viability was assessed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. MDA-MB-231 cells were seeded in 200 µl of culture media (full DMEM) per well in a 96-well plate and incubated 24 h prior to transfection with miR-17-5p mimics (according to the HiPerfect protocol; Qiagen GmbH) or treatment with the natural compounds. At 48 h post-transfection/treatment, 20 µl MTT solution (Sigma Aldrich, Germany) (5 mg/ml MTT in phosphate-buffered saline) was added to each well. Subsequent to incubation for 5 h, the formed formazan (MTT metabolic product) was re-suspended in 200 µl dimethyl sulfoxide (Sigma Aldrich, Germany). Colorimetric measurements and absorbance were performed using Wallac 1420 Victor 2 Multilabel Counter (Perkin Elmer Inc., Waltham, MA, USA) (Youness et al., 2016a). All cell viability experiments were performed in quadrates and repeated five times.

Cell migration was determined by wound healing assay post-transfection with miR-17-5p oligonucleotides or post-treatment with rosemary or Ursolic acid. Cells were seeded into a 6-well plate. Upon reaching 80%–90% confluence, the cell layer was scratched with a 10 μl pipette tip and washed with culture medium twice and cultured again for up to 48 h. Finally, the scratch wounds were visualized and imaged under a microscope (ZEISS Axio Observer Z1, Jena, Germany) (Youness et al., 2016b). All experiments were performed in triplicates and repeated at least three times. Cells were photographed at 0 h and after 24 h of culture to record the wound width.

We utilized the colony formation assay to determine the TNBC cell proliferative ability post-transfection using miR-17-5p oligonucleotides or post-treatment with Ursolic acid. MDA-MB-231 cells were transfected with miR-17-5p mimics (Qiagen, Germany). The transfected cells were then allowed to grow in 6-well plates and maintained in medium containing 10% FBS, and the medium was replaced after 6–7 days. After 14 days, cells were fixed with 4% formaldehyde and stained with 0.1% crystal violet at room temperature for 30 min. Colonies were counted and compared (Youness et al., 2019). All experiments were performed in triplicates.

Data are expressed as mean ± standard error of the mean (SEM) of at least 3 independent experiments. The statistical analysis was done by using Student’s t-test and P value < 0.05 was considered significant. While, one-way analysis of variance (One-way ANOVA) was performed for the purpose of comparison of more than two groups. Data was analyzed using GraphPad Prism 8.2.1 software.

Based on bioinformatic analysis, miR-17-5p was selected as the candidate miRNA that has the potential to dually target TNF-α and IL-10 with high binding scores (Figs. 1a and 1b, respectively). Different online algorithms were used. After the search was executed, results were sorted according to their mirSVR score. mirSVR is a regression model that computes a weighted sum of a number of sequence and context features of the predicted miRNA-mRNA duplex. Using miRbase.org database (http://www.mirbase.org), miR-17-5p accession number and mature sequence were retrieved (Table 2). In an attempt to investigate its relationship with other ncRNAs by bioinformatic analysis, miR-17-5p was found to hit the lncRNAs MALAT1 and H19 3’UTR sequence with high scores (Figs. 1c and 1d, respectively). It is also worth mentioning that the binding of miR-17-5p to H19 and MALAT have been experimentally validated in other studies (Jia et al., 2019; Xu et al., 2019b).

To study the expression of TNF-α and IL-10, we first evaluated both cytokines expression in 40 BC tumor samples using qRT-PCR. The expression levels of TNF-α and IL-10 were higher in BC tissues compared with their normal BC samples counterparts (P < 0.0001, Fig. 2a), (P = 0.0001, Fig. 2b), respectively.

To screen for the ncRNA transcription levels in BC, miR-17-5p, MALAT1 and H19 expression levels were evaluated in tumor breast tissues of the 40 BC patients. miR-17-5p was significantly down-regulated in BC tissues compared to normal tissues (P = 0.00930, Fig. 2c). On the other hand, MALAT1 and H19 levels were significantly upregulated in BC tissues compared to their normal counterparts (P = 0.0009, P = 0.0079, Figs. 2d and 2e).

Since literature concerning miR-17-5p role in cancer was controversial, it was essential to examine its impact in BC. Several BC hallmarks were assessed. Ectopic expression of miR-17-5p in MDA-MB-231 led to a significant decrease in cellular viability (P < 0.0001, Fig. 3a), clonogenicity (P = 0.0003, Fig. 3b) and cellular migration capacity of MDA-MB-231 cells compared to mock cells (P = 0.0433, Fig. 3c).

Efficient transfection of miR-17-5p oligonucleotides was successfully confirmed in MDA-MB-231 cells (P = 0.0062, Fig. 4a), which resulted in a significant decrease in TNF-α and IL-10 transcripts (P = 0.0080 and P = 0.0054, Figs. 4b and 4c) and protein levels (P < 0.0001 and P = 0.0001, respectively, Figs. 4d and 4e) compared to mock cells.

As noticed with TNF-α and IL-10 transcripts, miR-17-5p mimic resulted in a significant decrease in MALAT1 (P = 0.0002, Fig. 5a), and H19 (P = 0.0009, Fig. 5b) expression levels compared to mock cells.

To study the relationship between lncRNA MALAT1 and miR-17-5p target genes; TNF-α and IL-10, cells were transfected with MALAT siRNAs. Efficient knockdown of MALAT1 was first confirmed (P = 0.0223, Fig. 6a), then the influence of MALAT1 silencing on the cytokines levels was investigated. A significant decrease in TNF-α (P = 0.0263, Fig. 6b) and IL-10 (P = 0.0057, Fig. 6c) transcript levels compared to their respective mock cells was observed.

Similarly, the effect of lncH19 knocking down was investigated. Efficient knockdown of H19 was first confirmed using qPCR (P = 0.0017, Fig. 7a), which caused a significant decrease in TNF-α (P < 0.0001, Fig. 7b), and IL-10 (P = 0.0032, Fig. 7c) transcript levels.

To consolidate the mutual relationship between MALAT1 and H19, H19 transcript levels were assessed in response to MALAT1 siRNAs transfection. It was noticeable, that Knocking down of MALAT1 in the MDA-MB-231 cells resulted in a significant decrease in lncRNA H19 transcript levels (P = 0.0072, Fig. 8a). On the other hand knocking down of MALAT-1 resulted in a significant increase in miR-17-5p levels (P = 0.0033, Fig. 8b) confirming their mutual relationship.

Similar results to those of MALAT1 siRNAs transfection was observed with H19 siRNAs; where H19 knocking down resulted in a significant decrease in lncRNA MALAT1 (P = 0.0488, Fig. 9a) and increase in miR-17-5p transcript levels (P < 0.0001, Fig. 9b).

Further approach was taken to investigate the ability of natural compounds to regulate ncRNAs and their respective target cytokines. Rosemary is known for its potential anticancer effect. Fractionation of rosemary aerial parts extract was first performed using dichloromethane and ethyl acetate (EA). Cellular viability of MDA-MB-231 in response to rosemary fractions exposure was evaluated relative to vehicle control cells. Both fractions led to significant decrease in cellular viability, either with diff concentrations (30μM-60μM) of dichloromethane fraction (P ≤ 0.0001, Fig. 10a) or EA fraction (P ≤ 0.0001, Fig. 10b).

HPLC-PDA-ESI-MS was used to characterize phenolic metabolites within the bioactive fraction Fig. S1). Metabolites profiling revealed the presence of the phenolic acid, ursolic acid, as the major peak with peak area of 2.5%. The MS, 1H-NMR and 13C-NMR spectral data was compared closely with that of the reported by the literature (Annan et al., 2011).

In these experiments, UA was chosen as a major active constituent of the rosemary extract. To functionally analyze the UA effect on TNBC cell lines, MDA-MB-231 cells were treated with different concentrations of UA (20 μM–100 μM). MTT assay was performed and cellular viability was calculated relative to mock cells. A significant decrease in cellular viability was observed upon UA treatment compared to vehicle control (P < 0.0001, Fig. 10c). To investigate the long-term effects on MDA-MB-231 TNBC cell lines, colony forming assay was performed. UA led to a dose-dependent decrease in number of colonies reaching to significance only with the highest concentration (100 μM) (P = 0.0358, Fig. 10d). Cellular migration was calculated as percentage closure of wound. UA treatment resulted in a marked decrease in cellular migration of MDA-MB-231 cells (P = 0.0414, Fig. 10e).

MDA-MB-231 cells were treated with 60 µM UA then miR-17-5p, MALAT1 and H19 expression levels were assessed using qRT-PCR. Results showed a significant increase in miR-17-5p (P < 0.0001, Fig. 11a), a significant increase in MALAT1 (P < 0.0001, Fig. 11b), and a significant decrease in H19 (P = 0.0094, Fig. 11c) expression levels. For cytokines, TNF-α expression showed an increase (P = 0.0003, Fig. 11d) while IL-10 expression was decreased (P < 0.0001, Fig. 11e) upon UA treatment.