Patient cohorts and in silico analysis

To assess the prognostic significance of miR-4649-5p and PIP5K1C in breast cancer patients, an online publicly available Kaplan–Meier plotter was used that provides separate specialized software tools for analyzing the impact of microRNAs, mRNA, or protein on the survival of breast cancer patient cohorts originating from different databases (https://kmplot.com/analysis/index.php?p=service&cancer=breast_mirna [23] for miR-4649-5p; https://kmplot.com/analysis/index.php?p=service&cancer=breast [24] for PIP5K1C on the RNA level, and https://kmplot.com/analysis/index.php?p=service&cancer=breast_protein [25] for PIP5K1C on the protein level). For miR-4649-5p, data on the overall survival of 97 patients with triple negative breast cancer (TNBC) was available from The Cancer Genome Atlas Project (TCGA) [23]. For PIP5K1C the overall survival of 153 TNBC patients as well as the relapse-free survival of 392 TNBC patients was analyzed based on gene chip data originating from the Gene Expression Omnibus (GEO) database [24], and the overall survival of 65 breast cancer patients without subtype discrimination was analyzed based on proteomics data [25]. No subtype discrimination could be applied in this case, as the cohort size was too small to limit it to TNBC patients. For all survival analyses, a follow-up threshold of 60 months was set and patients were split into high and low expression groups by an auto-selected optimal cutoff. The online analysis tool TNMplot [26] was employed to analyze the expression of PIP5K1C in invasive breast carcinomas (n = 112) compared to paired healthy tissue (n = 112) from RNA-seq-based TCGA data.

Cell lines and cell culture conditions

The following triple negative breast cancer cell lines were used in this study: SUM159, MDA-MB-231, and BT-20. MDA-MB-231 and BT-20 cells were obtained from the American Type Culture Collection (ATCC; Manassas, CA, USA). SUM159 cells were purchased from Asterand (Detroit, MI, USA). Furthermore, HEK239 cells purchased from the American Type Culture Collection (ATCC) were used.

MDA-MB-231 and HEK239 cells were maintained in high-glucose DMEM (4.5 g/l d-Glucose, l-Glutamine, 25 mM HEPES; Gibco, Thermo Fisher Scientific, Waltham, MA, USA), 10% fetal bovine serum (FBS) (Serana, Pessin, Germany), and 1% penicillin/streptomycin (final concentration penicillin: 100 units/ml, final concentration streptomycin: 100 µg/ml; Sigma-Aldrich, St. Louis, MO, USA). SUM159 cells were grown in Ham's Nutrient Mixture F12 containing 1 mM l-Glutamine (GE Healthcare Life Sciences, Pittsburgh, USA) supplemented with 2 mM HEPES buffer (Gibco), 5 μg/ml insulin (Sigma-Aldrich), 1 μg/ml hydrocortisone (Sigma-Aldrich), 1% penicillin/streptomycin (Sigma-Aldrich), and 5% FBS (Serana). BT-20 were cultivated in Minimum Essential Medium—Eagle with Earle's BSS and l-Glutamine (Lonza, Basel, Switzerland) supplemented with 10% FBS (Serana) and 1% penicillin/streptomycin (Sigma-Aldrich). All cell lines were cultivated in a 5% CO2 humidified incubator at 37 °C. SUM159 and MDA-MB-231 cells were tested for mycoplasma with the Venor GeM Mycoplasma Detection Kit (Minerva Biolabs, Berlin, Germany) by the Core Facility for Alternative Biomodels and Preclinical Imaging of the Medical University of Graz, Austria.

Transient miR-4649-5p mimic and PIP5K1C siRNA transfection

To achieve transient overexpression of miR-4649-5p or knockdown of PIP5K1C, TNBC cell lines were transfected with 10 nM mirVana™ hsa-miR-4649-5p mimic or mirVana™ mimic control (Thermo Fisher Scientific) and 20 nM PIP5K1C siRNA #8 (targeting sequence: TTCCTGTACTGTAAAGACTAA; Qiagen) or AllStars Negative Control siRNA (Qiagen), respectively.

Cells in 6-well plates or 6 cm dishes were transfected using the HiPerFect Transfection Reagent (Qiagen) following the fast-forward transfection protocol of the manufacturer. Cells in 96-well plates were transfected according to the reverse transfection protocol. To confirm efficient overexpression or knockdown 48 h after transient transfection, quantitative RT-PCR was applied.

Generation of stable miR-4649-5p mimic cells by lentiviral transduction

SUM159 cells were seeded and incubated overnight in complete growth medium in a 24-well plate. After 24 h, medium was replaced with complete growth medium containing ViralPlus Transduction Enhancer (ABM, Richmond, BC, Canada) diluted 1:200 and 10 μg/ml polybrene (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Cells were transduced by dropwise addition of 10 μl of miR-4649-5p mimic virus (LentimiRa-GFP-hsa-miR-4649-5p Virus, ABM) or control virus (Lenti-III-mir-Off Control Virus, ABM). 48 h after transduction, cells were selected with 1.5 μg/ml puromycin dihydrochloride (Gibco) for 1 week while monitoring GFP expression by fluorescent microscopy. Subsequently, cells were sorted by FACS for high GFP expression. Overexpression of miR-4649-5p was checked by quantitative RT-PCR.

RNA isolation and cDNA synthesis for quantitative RT-PCR

RNA was isolated from cell lines at a confluency of 75–95% in biological triplicates using TRIzol™ Reagent (Thermo Fisher Scientific) following the manufacturer`s protocol. For quantitative RT-PCR, 1 μg of total RNA per sample was reverse transcribed into cDNA either with the miScript II RT Kit (Qiagen) using the miScript HiFlex Buffer for the simultaneous detection of miRNA and mRNA, or the QuantiTect Reverse Transcription Kit (Qiagen) for the detection of only mRNA, both according to the manufacturer’s protocols.

Quantitative RT-PCR (RT-qPCR)

Quantitative RT-PCR was carried out in technical duplicates using the QuantiTect SYBR Green PCR Kit (Qiagen) according to the manufacturer’s two-step RT-PCR protocol. To measure miR-4649-5p expression, the Hs_miR-4649-5p_1 miScript Primer Assay (Qiagen) was used together with the miScript Universal Primer from the miScript SYBR Green PCR Kit (Qiagen) and was normalized to the two housekeepers SNORD61 and SNORD95 using the Hs_SNORD61_11 and Hs_SNORD95_11 miScript Primer Assays (Qiagen) according to the manufacturer`s instructions.

For the detection of coding genes together with the housekeepers GAPDH and U6, primers were designed with the NIH Primer Blast Tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) and purchased from Eurofins Scientific. A list of these primer sequences is given in Additional file 1: Table S1.

Per qPCR reaction (volume of 10 µl), 1 ng of cDNA was used. The measurements were carried out in LightCycler® 480 Multiwell Plates 384 on a LightCycler® 480 Real-Time PCR System (Roche, Basel, Switzerland). Ct values were normalized by subtracting the respective arithmetic mean of the two housekeeper genes to receive delta Ct (ΔCt) values. Following the ΔΔCt method, relative expression levels were calculated by subtracting the ΔCt of the respective negative control and were plotted as 2−ΔΔCt.

Digital droplet PCR (ddPCR)

To determine miR-4649-5p expression by ddPCR, RNA was isolated using TRIzol™ (Thermo Fisher Scientific) as indicated by the manufacturer and 10 ng total RNA were reverse transcribed with the TaqMan Advanced miRNA cDNA Synthesis Kit (Thermo Fisher Scientific) according to the protocol. cDNA was diluted 1:100 and 5 μl were used per 20 μl reaction with 10 μl ddPCR Supermix for Probes (No dUTP; BioRad) and 0.5 μl TaqMan Advanced miRNA Assay for miR-4649-5p (Thermo Fisher Scientific). Droplets were generated from the samples in duplicates with a QX200™ droplet generator (BioRad) using Droplet Generation Oil for Probes (BioRad) and DG8™ Cartridges for QX200™ (BioRad) according to the manufacturer. Droplets were transferred into semi-skirted 96-well ddPCR plates (BioRad), plates were heat-sealed, and PCR was performed on a T100TM Thermal Cycler (BioRad) (95 °C 10 min; 95 °C 30 s + 61 °C 1 min × 39; 98 °C 10 min; ramping 2 °C/s). The droplets were read on a QX200™ droplet reader device (BioRad) and data were evaluated using the QX Manager Software Version 1.2 (BioRad).

WST-1 cell growth assay

To assess the impact of miR-4649-5p and PIP5K1C on cellular growth, 3 × 103 SUM159, 5 × 103 MDA-MB-231, and 6 × 103 BT-20 cells were seeded per well of a 96-well plate (one plate for each time point) and transiently transfected with the miRNA mimic, PIP5K1C siRNA and their respective controls in six replicates using HiPerFect Transfection Reagent (Qiagen) according to the reverse transfection protocol of the manufacturer. To investigate potential additive effects of miR-4649-5p mimic and PIP5K1C or AKT inhibition, cells were additionally treated with 10 μM of the PIP5K1C inhibitor UNC3230 (MedChemExpress, Monmouth Junction, NJ, USA), 0.5 μM of the AKT inhibitor capivasertib (MedChemExpress) or equal volumes of DMSO as vehicle control. Cells were incubated for 24, 48, 72, and 96 h. At each time point, WST-1 proliferation reagent (Roche) was added to the wells in a 1:10 ratio according to the manufacturer’s instructions. After incubation at 37 °C for 60 or 120 min (depending on the cell line), colorimetric changes were measured using a SPECTROstar Omega (BMG Labtech, Ortenberg, Germany) at a wavelength of 450 nm with a reference wavelength of 620 nm.

Colony formation assay

To confirm the impact of altered miR-4649-5p and PIP5K1C expression on cell growth over a longer time frame, we performed clonogenic assays. Transiently transfected cells were trypsinized 24 h after transfection, counted, and seeded in 6-well plates with 200 (SUM159) or 500 (MDA-MB-231) cells/well. Cells were cultured at 37 °C and 5% CO2. After 7 (SUM159) or 14 days (MDA-MB-231), cells were fixed and stained with 0.04% crystal violet (Sigma-Aldrich) in 20% methanol/PBS. The number of colonies was counted macroscopically. Each experiment was carried out in triplicates or sextuplicates.

Protein extraction and western blotting

Total proteins were isolated from fresh cells using Radioimmunoprecipitation Assay (RIPA) buffer (Sigma-Aldrich) supplemented with 1:50 protease inhibitor cocktail P8340 (Sigma-Aldrich). From each sample, 25 µg of protein were diluted in 4× Laemmli buffer (BioRad, Hercules, CA, USA) containing 10% β-mercaptoethanol (Sigma-Aldrich) and separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) on a 4–15% Mini-PROTEAN® TGX™ Precast Gel (BioRad) before plotting on nitrocellulose membranes (BioRad). Membranes were blocked for at least 1 h with 5% non-fat milk powder in 1× Tris-buffered saline (TBS; BioRad)/0.1% Tween-20 (Sigma-Aldrich) before incubation with the primary antibodies overnight at 4 °C. Primary antibodies were diluted in 5% BSA (Sigma-Aldrich)/TBS-Tween (BioRad). The following primary antibodies were used: PIP5K1C (#3296, Cell Signaling Technology, Cambridge, United Kingdom) diluted 1:1000, AKT (#9272, Cell Signaling Technology) diluted 1:1000, Phospho-Akt (S473, D9E, #4060, Cell Signaling Technology) diluted 1:2000, and the housekeeper Cofilin (ab42824, Abcam, Cambridge, United Kingdom) diluted 1:10,000. After primary antibody incubation, membranes were washed (three times for 10 min each in TBS-Tween), incubated with the secondary antibody (HRP-conjugated anti-rabbit; diluted 1:1000 in 5% milk/TBS-Tween; Santa Cruz Biotechnology, Dallas, TX, USA) for 1 h and washed again. Signals were detected with an enhanced chemiluminescence detection system (SuperSignal™ West Pico PLUS Chemiluminescent Substrate, Thermo Fisher Scientific) on a BioRad ChemiDoc Touch device. Densitometric quantifications were performed using the Image Lab Software (BioRad). For reprobing, membranes were stripped with 10% acetic acid for 1 h.

EdU proliferation assay

Cell proliferation was assessed in a flow cytometric assay using the Click-iT™ Plus EdU Pacific Blue™ Flow Cytometry Assay Kit (Thermo Fisher Scientific). Cells were seeded and transfected in 6 cm dishes using HiPerfect (Qiagen) as stated previously. Cells were labeled with 10 µM EdU in full growth medium for 2 h at 37 °C and 5% CO2. All subsequent steps were performed according to the Click-iT™ Plus EdU Pacific Blue™ Flow Cytometry Assay protocol. The samples were measured on a CytoFLEX SI (Beckman Coulter) recording 20,000 events per sample. Cell populations were gated to exclude debris/dead cells (R1) and cell aggregates (R2).

Caspase 3/7 activity assay

To analyze the induction of apoptosis, the activity of effector caspase-3 and -7 was measured using a Caspase-Glo® 3/7 assay (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Cells were transiently transfected in four technical replicates in 96-well plates with 3 × 103 SUM159 and 5 × 103 MDA-MB-231 cells per well using HiPerFect Transfection Reagent (Qiagen) according to the reverse transfection protocol. After 48 h, the luminogenic reagent was added to the cells as instructed in the protocol, and signals were measured with a LUMIstar Omega (BMG LabTech).

Cell migration assays

Two types of cell migration assays were conducted. First, wound healing assays, also called scratch assays, for which SUM159 cells were seeded and transiently transfected in 6-well plates in four biological replicates. To reach confluence, high cell numbers were seeded (5 × 105) and scratches were introduced after 24 h. Medium was changed and cells were washed to remove scratched-off cells. Closure of the scratches was observed under the microscope after 15/20 h, 24 h, and 45 h. The area of the scratches was determined using the ImageJ (NIH, Bethesda, Maryland) plugin “MRI Wound Healing Tool” and for each time point, the remaining area relative to the 0 h time point was calculated.

To confirm the results of the wound healing assays, transwell migration assays were performed using transwell inserts with 0.4 µm pore size (Corning Incorporated, Corning, NY, USA). Cells were transiently transfected in 6-well plates and after 24 h medium was changed to start starvation without FBS. After 24 h of starvation, cells were seeded on transwell membranes (1.5 × 104 cells for SUM159, 2.5 × 104 cells for MDA-MB-231). The membranes were previously coated with 0.1% gelatin (Sigma-Aldrich) in 0.02 M acetic acid, left to dry overnight, and rehydrated with FBS-free medium at 37 °C for 1 h before seeding the cells in FBS-free medium. The lower chambers were filled with full growth medium containing FBS. After 48 h, cells were fixed with cold methanol and stained with 0.2% crystal violet (Sigma-Aldrich) in 2% ethanol. After washing, cells that did not migrate through the membrane were removed. From each transwell, microscopic images of five representative areas were taken at 40× magnification and counted manually.

Transcriptome analysis and identification of potential miR-4649-5p targets

For whole transcriptome analysis by RNA-seq, total RNA was isolated from fresh cell pellets with the RNeasy Mini Kit with DNAse treatment according to the manufacturer’s instructions (Qiagen). After quality control of the RNA with an Agilent RNA 6000 Nano Kit on an Agilent 2100 BioAnalyzer system (Agilent Technologies, Santa Clara, CA, USA) and quantification on a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific), 250 ng of total RNA were used for library preparation with the NEBNext® rRNA Depletion Kit v2 in combination with the NEBNext® Ultra™ II Directional RNA Library Prep Kit for Illumina® (New England Biolabs GmbH, Frankfurt am Main, Germany) according to manufacturer’s instructions. Libraries were QC checked with an Agilent 2100 DNA high sensitivity kit, pooled, and sent to Vienna BioCenter Core Facilities GmbH (Vienna, Austria) for sequencing in an Illumina NovaSeq SP flow cell (Illumina, Eindhoven, Netherlands) in SR100 mode. After demultiplexing, FASTQ files were used for data analysis.

To identify potential miRNA targets, significantly downregulated genes were determined based on p values adjusted for multiple testing and ranked according to their log2-fold change values. In silico target predictions were performed for the top candidates using the online tools TargetScan [27], miRWalk2.0 [28], and miRDB [29].

A Gene Ontology (GO) enrichment analysis was performed on significantly downregulated genes using the PANTHER Overrepresentation Test (http://www.pantherdb.org/; GO Ontology database https://doi.org/10.5281/zenodo.6799722, released 2022-07-01, reference list Homo sapiens, annotation data set GO biological process complete, Fisher’s Exact Test).

Dual luciferase reporter assay

To confirm the direct interaction of miR-4649-5p with the putative target PIP5K1C, a 60 nt region of the 3′ UTR of PIP5K1C containing a predicted binding site for the miRNA was inserted into the dual luciferase reporter vector pEZX-MT06 (Genecopoeia, Rockville, MD, USA). Both the wild-type (5′ CCCCAAACACTGGTTTGCATCCCAGGTTCCTCGCCCACCTACCCCCGCCACACCCCGTCT 3′) and a mutated binding site sequence (5′ CCCCAAACACTGGTTTGCATCGTAGGTTCCAGGTTCACCTACCCCCGCCACACCCCGTCT 3′) were used. An empty control plasmid (CmiT000001-MT06; Genecopoeia) was employed as a reference control. For the luciferase assay, HEK293 cells were seeded in 24-well plates. Once cells reached 70–80% confluence they were co-transfected with 200 ng pEZ-MT06 PIP5K1C wt/mutated reporter vector or control vector and 50 nM mirVana™ miR-4649-5p mimic or mirVana™ mimic control (Thermo Fisher Scientific) using Lipofectamine 2000 Transfection Reagent (Thermo Fisher Scientific) and Opti-MEM Reduced Serum Medium (Thermo Fisher Scientific) according to the manufacturer’s instructions. Cells were harvested 24 h after transfection and the Luc-Pair Luciferase Assay Kit 2.0 (Genecopoeia) was performed according to the user manual. Luminescence was measured with a LUMIStar Omega luminometer (BMG LabTech). The firefly luciferase signals were normalized by the renilla luciferase signals.

Statistical analysis

Statistical analysis was performed using GraphPad Prism Version 5.01 (GraphPad Software, Inc., San Diego, CA, USA). Differences between transfected samples and respective controls were assessed by unpaired, two-tailed independent t-tests with a 95% confidence interval, unless indicated otherwise. p values below 0.05 were considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.005).