The TNBC cell line MDA-MB-231 was purchased by ATCC (Rockville, USA) and their brain-seeking subclone MDA-MB-231BR was a kind gift from Prof. Dr. Wikman-Kocher (Institute of Tumor Biology, University Medical Center Hamburg-Eppendorf (UKE), Hamburg, Germany). All cell lines were cultured in DMEM-medium (Gibco, ThermoFisher Scientific, Waltham, MA USA) supplemented with 10% (v/v) fetal calf serum (FCS) under standard conditions in a water-saturated atmosphere containing 5% CO₂ at 37°C. Ipatasertib (GDC-0068) was purchased from Selleckchem (Cat.-No. S2808, Biozol, Eching, Germany, dissolved in Dimethylsulfoxide (DMSO)) and used in concentrations of 1 to 20 µM.

AKT1 knockout (AKT_KO) in MDA-MB-231BR cell line was performed using the protocol described by Ran et al. (20). Two guide sequences of the AKT1 gene, located on exon 3 (guide #1, guide sequence: GAGCGACGTGGCTATTGTGA) and exon 4 (guide #5, guide sequence: TGGCTACAAGGAGCGGCCGC) were generated using CRISPR guide RNA design tool guide scan. Primers used for sgRNA oligo insert construction can be found in the Supplementary Materials and Methods part. The designed sgRNA oligos were used for cloning into the BbsI-cleaved pSpCas9(BB)-2A-Puro(PX459)-GFP plasmid (kindly provided by PD Dr. Stefan Werner, Institute of Tumor Biology, UKE). MDA-MB-231BR cells were seeded in 6-well-plates (3.0 x 10⁵/well) and after 24h, cells were transfected with the CRISPR/Cas9-plasmids AKT1_e3_guide#1, AKT1_e4_guide#5 or AKT1_empty vector (Ctrl) using Lipofectamine LTX (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA). Recombinant cells were passaged in DMEM-medium containing 1 µg/mL puromycin (PAA Laboratories GmbH, Pasching, Austria). The transfection efficiency was confirmed assessed by fluorescence microscopy. Transfected single cell colonies were obtained by FACS Sorting method (BD FACSAria™ III sorter, Becton Dickinson, Franklin Lakes, New Jersey, USA). The success of AKT1 knockout in MDA-MB-231BR cells was confirmed by Sanger-sequencing (Eurofins Genomics GmbH, Ebersberg, Germany). The mutations of AKT1 were analyzed using the CRISP-ID tool (21). AKT1-protein expression level was further verified using Western Blot. In the end, two AKT1_KO clones containing the e4_guide#5 (AKT1_KO_C1 and AKT_KO_G4) were expanded and used for the further in vitro-experiments.

Cells were seeded in 6-well-plates (2.5 x 10⁵/well) and allowed to adhere for 24h. Whole-cell lysates were prepared using RIPA lysis-buffer containing protease inhibitor (1x, Sigma-Aldrich, St. Louis, Missouri, USA) and phosphatase inhibitor (1x, Merck Millipore, Darmstadt, Germany) and analyzed by Western blot as previously described (22). The primary antibody rabbit anti-AKT1 (#2938, Cell Signaling Technology) was used for specific AKT1-detection and the mouse anti-β-Actin (sc-47778, Santa Cruz Biotechnology) was used as loading control. Immunocomplexes were detected using peroxidase-conjugated mouse anti-rabbit IgG (sc-2357, 1:8000, Santa Cruz Biotechnology) and goat anti-mouse IgG (1030-05, 1:8000, Southern Biotech) followed by chemiluminescence detection (Westar Nova 2.0, Cyanagen Srl, Bologna, Italy) and Fuji Medical X-Ray Film (Fuji-film Corporation).

Cells (MDA-MB-231 or MDA-MB-231BR, 5000/well) were seeded in quintuplicates in 96-well-plates, grown for 24h and treated with Ipatasertib (1, 5, 10 or 20µM) or 0.1% DMSO as control by replacing the cell culture medium for 24, 48 and 72h. MDA-MB-231BR AKT1_KO or Ctrl cells were seeded in quintuplicates as described above in cell culture serum-reduced medium and grown for 24, 48 and 72h. Cell viability was determined by Cell Proliferation Reagent WST-1 assay (Roche, Mannheim, Germany) using the manufacturer’s protocol and Microplate Absorbance Reader (Tecan Trading AG, Schweiz). Cell viability values of each cell line after 2 hours were used for normalization. For statistical analyses, each experiment was performed three-times (n=3) and results were given as mean ± S.D.

Cells (2.3 x 10⁵/well) were seeded in duplicates in 12-well-plates and after achieving full confluency, artificial scratches were performed using a sterile 200µl micropipette tip. Cells were washed with Phosphate-Buffered Saline (PBS), containing Ca²⁺ and Mg²⁺ (+/+), and Ipatasertib (1, 2.5, 5, 10, 20 µM) was added. Images of the gaps (two images/area/well) were taken always at the same place every 6h under the light microscope (Zeiss, Oberkochen, Germany). The evaluation was performed using ImageJ Wound Healing Tool (Wayne Rasband, National Institute of Health). For statistical analyses, experiment was performed three-times (n=3) and results were given as mean ± S.D.

To establish the Ipatasertib IC₂₀-values of MDA-MB-231 and MDA-MB-231BR, cells were seeded in sextuplicates in 6-well-plates (250/well), grown for 24h and incubated with Ipatasertib (1-20µM) or 0.1% DMSO. After 24h, cells were washed with PBS and new cell culture medium was added for additional 10d allowing colony growth. Established colonies were fixed with 70% ethanol and stained with 1% crystal violet (Sigma-Aldrich). For IC₂₀-calculation, numbers of colonies (colony≥50 cells) were counted manually under the light microscope. To analyze the effect of AKT-inhibition on radiation, cells were plated as described above and grown for 24h. Ipatasertib (IC₈₀-value) or 0.1% DMSO was added followed by 2, 4, 6Gy radiation 2h later. No-radiated cells were used as control. After 24h, cell culture medium was changed and cells were incubated for additional 9 to 10d. Colonies were fixed and stained as described before, and colony growth was evaluated (23). For statistical analyses, experiment was performed two-times (n=2) and results were given as mean ± S.D.

Cell migration assay was performed in Boyden’s transwell chambers with 8.0µM pore size filters (Corning, NY, USA). Cells (5.000/200µL/well) were placed in triplicates in the upper side of Boyden’s chamber and were treated with 20µM Ipatasertib or 0.1% DMSO as control. In the first setting, cell suspension in cell culture medium with 5% FCS was placed in the upper side, whereas the lower chamber contained cell free culture medium with 10% FCS. In the second setting, conditioned-medium was used from astrocytes, which were previously cultured 72h in 3% FCS containing DMEM-medium. Tumor cells were cultured in conditioned-medium and were placed in the upper side of the chamber and in the lower side, cell free culture medium containing 10% FCS was added. In both setting, inserts containing cells were fixed after 24h in 4% formaldehyde and stained with 1% crystal violet. Then, cells from the upper side of the Boyden’s chamber, which didn’t migrate through the pores, were removed by wiping with a cotton swab. The stained membranes were cut out from the inserts and fixed on an object carrier using Eukitt mounting medium (Sigma-Aldrich), covered by a cover glass and assessed under the light microscope. Cells were counted manually. For statistical analyses, experiment was performed two-times (n=2) and results were given as mean ± S.D.

To detect genomic variants in AKT1_KO MDA-MB-231BR cells, whole-genome sequencing (WGS) was performed. AKT1_KO_C1, AKT1_KO_G4 and Ctrl cells were seeded in 6-well-plates (2.5 x 10⁵/well) and were grown to 80% cell confluency. DNA was extracted using a QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany) and quantity and quality were verified by NanoDrop and gel electrophoresis with 1% agarose gel, respectively. PCR free Library construction and Sequencing (DNBseq PE150) was performed by BGI Genomics (Shenzhen, China). Further analysis of the data set can be found in Supplementary Materials and Methods part.

To identify differentially expressed genes (DEGs) related to AKT-dependent cell migration, invasion and radiosensitivity, RNA sequencing (RNAseq) was performed. MDA-MB-231BR AKT1_KO clone C1, clone G4 and Ctrl cells were seeded in 6-well-plates (2,5 x 10⁵/well) in quadruplicates and grown to 80% cell confluency. Cells were washed with PBS (+/+) and total RNA, treated by DNase, was isolated by using Qiagen Plasmid Midi Kit and the manufacturer’s protocol (Qiagen). Sample purity was assessed using Agilent Bioanalyzer RNA 6000 Nano Kit (Agilent Technologies Inc., CA, USA) and appropriated samples (RNA 28S/18S ≥ 1.0 and RIN ≥ 7.0) were used for transcriptome analysis. RNAseq library preparation (non-stranded mRNA enrichment method) with DNBseq PE100 sequencing were performed by BGI Genomics (Shenzhen, China). Further analysis of the data set can be found in Supplementary Materials and Methods part.

For the functional assays, statistical analyses of the data were performed with GraphPad Prism Software 8.0 (San Diego, CA, USA) and comparison among groups were made by one-way or two-way ANOVA with Bonferroni post hoc tests. Differences were considered statistically significant at a p-level less than 0.05.

To investigate the role of the serine/threonine kinase AKT1 on the cellular function of BCBM cells, we initially used the potent pan-AKT inhibitor Ipatasertib and the brain-seeking TNBC cell line MDA-MB-231BR for functional analysis. The cytotoxic effect of Ipatasertib at different concentrations (1, 5, 10, 20µM) was evaluated using the WST-1 assay. Here, Ipatasertib-treatment in a low therapeutic dose resulted in a slight reduction of the MDA-MB-231BR cell viability, whereas the higher concentration of 20µM showed a significant decrease in cell viability ( Figure 1A ). Precisely, 1µM Ipatasertib reduced the amount of vital cells to 90.50% after 72h and 20µM Ipatasertib to 53.30% (p=0.001), respectively. Besides this data, the parental cell line MDA-MB-231 showed no changes in the cell viability with the indicated Ipatasertib-concentrations. Next, we evaluated the impact of AKT-inhibition on MDA-MBA-231 BR cell migration. Therefore, we performed a wound healing-assay by scratching a cell-less area after 100% confluency of MDA-MB-231BR cells and treated them for 24h with Ipatasertib. In contrast to the cell viability assays, no change in cell migration of MDA-MB-231BR cells was observed under treatment with different Ipatasertib-concentrations after 6, 12 and 24h ( Figure 1B ). As shown in the Figure 1C , wounds were closed within 24h in all conditions ( Figure 1C ). To further analyze the effect of Ipatasertib on MDA-MB-231BR cell migration, Boyden`s transwell-assays were performed. AKT-inhibition resulted in no significant alteration of cell migration compared to MDA-MB-231BR untreated cells ( Figure 1D left). In a next step, we asked whether PTEN-inhibition, via culturing the tumor cells with astrocytes-conditioned medium (astrocytes-CM), might have an impact on AKT-mediated MDA-MB-231BR cell migration and in turns on AKT-inhibition. The data demonstrated, that the astrocytes-stimulated untreated MDA-MB-231BR cells had the same values of migrated cells compared to Ipatasertib-treated MDA-MB-231BR cells ( Figure 1D right). Furthermore, 20µM Ipatasertib-treatment had no significant impact on PTEN-mediated cell invasion compared to untreated control cells.

In conclusion, our findings indicated that the pan-AKT inhibitor, solely in high concentration, had a significant impact on the proliferation of MDA-MB-231BR cells. In contrast, AKT-inhibition had no effects on the brain seeking MDA-MB-231 cell migration, neither in the absence nor presence of astrocytes-mediated stimulating loss of PTEN.

Next, we aimed to analyze whether the pan-AKT inhibitor Ipatasertib might sensitize BC cells for irradiation. For this purpose, we used the MDA-MB-231BR as well as the parental cell line and determined their IC₂₀-value, which represents the Ipatasertib-concentration leading to 20% clonogenic survival inhibition. As a result, we obtained different IC₂₀-values for both cell lines and found that the MDA-MB-231BR cell line respond more sensitively (IC₂₀= 2.5µM) to Ipatasertib-treatment than the parental cell line (IC₂₀= 6µM) ( Supplementary Figures 2A, B ). In the next step, both cell lines were treated with Ipatasertib for 2h using the established IC₂₀-values, and subsequently irradiated with 2, 4 and 6Gy. After 10 days, cell colonies were stained with crystal violet and counted. MDA-MB-231 showed a dose-dependent reduction in cellular survival with no change after Ipatasertib-treatment ( Figure 2A ). In contrast, AKT-inhibition in the MDA-MB-231BR resulted in a significant radiation sensitization that reached statistically significant at a dose of 6Gy radiation dose (p=0.04) ( Figure 2B ).

Our results indicated, that Ipatasertib-treatment at IC₂₀-value sensitize MDA-MB-231BR cells to therapeutic irradiation at higher doses.

Since Ipatasertib is a pan-AKT inhibitor and we were interested in the specific role of the AKT1 isoform, we next established MDA-MB-231BR sublines with an AKT1 knockout. Here, two different clones, AKT1_KO_C1 and AKT1_KO_G4, and corresponding Ctrl were generated by CRISPR/Cas9-system. The Sanger-sequencing analyses revealed that in both clones the two AKT1-alleles displayed different genomic alterations and were consequently heterozygous variants ( Figure 3A ). AKT1_KO_C1 clone showed a deletion of one nucleotide on different positions at allele 1 and allele 2. The AKT1_KO_G4 clone showed on the allele 1 a deletion of two nucleotides and on allele 2 a deletion of seven nucleotides. Those results were in line with the WGS-data, which indicated, that both AKT1_KO clones were affected by frame-shift mutations and that the clone AKT1_KO_G4 showed a more pronounced somatic alteration compared to the clone AKT1_KO_C1 ( Supplementary Table 1 ).

At the mRNA-level, RNAseq data showed an effective AKT1-downregulation of both AKT1_KO clones compared to Ctrl ( Figure 3B ), whereas the isoforms AKT2 and AKT3 were not affected. Western blot analysis confirmed the effective AKT1 knockout in MDA-MB-231BR cells at protein level.

In line with mRNA level data, AKT2 and AKT3 were not affected by AKT1 knockout as Western blot analysis revealed isoform-specific protein-bands with slight differences in expression compared to Ctrl cells. As expected, AKT1 knockout led to an effective reduced AKT-phosphorylation in MDA-MB-231BR_C1- and _G4-clones ( Figure 3C ).

Our data revealed, that effective knockout of AKT1 leading to a loss of the isoform AKT1 in both MDA-MB-231BR established clones.

Next, we analyzed the functional role of AKT1 on cell proliferation. The ability of reducing cell viability was analysed in the two MDA-MB-231BR AKT1_KO clones and the values were compared to those of MDA-MB-231BR_Ctrl cells. While AKT1_KO_G4 cells behaved like the Ctrl cells, AKT1_KO_C1 cells showed a significant reduction (p=0.006) in the cell viability after 72h ( Figure 4A ). Furthermore, the impact of AKT1 knockout on tumor cell migration was evaluated using a transwell-assay with the aforementioned cell lines. Surprisingly, we observed an apparent increased cell migration in AKT1_KO clones compared to Ctrl cells ( Figure 4B ). AKT1 knockout led to faster migration of MDA-MB-231BR cells and the effect was statistically significant in AKT1_KO_C1 clone (p=0.0027) in comparison to Ctrl cells.

These results indicated that the loss of AKT1 contributes to a reduction of cell proliferation and the promotion of cell migration in MDA-MB-231BR cells.

To determine the impact on AKT1-mediated radiosensitization on BC cells, AKT1_KO clones as well as the corresponding Ctrl cells were irradiated with 2, 4 and 6Gy and the clonogenic survival was analyzed as previously described. It was shown that the knockout of AKT1 led to significant radiation resistance at a dose of 6Gy compared to the Ctrl cells, which was already indicated at a dose of 4Gy (AKT1_KO_C1 p=0.01). This trend towards higher radiation resistance was also evident in the second AKT1 knockout clone (AKT1_KO_G4), but did not reach a significant level ( Figure 4C ). Next, the clonogenic potential of the aforementioned cell lines was compared in the absence of treatment. Our results showed that the colony formation ability of AKT1_KO_C1 (p=0.0195) and AKT1_KO_G4 (p=0.04) was significantly increased compared with MDA-MB-231BR_Ctrl ( Figure 4D ).

In conclusion, AKT1 knockout increases the clonogenic potential of MDA-MB-231BR cells and in turn leads to an enhanced radioresistance.

To further elucidate the underlying molecular mechanism of the unexpected increased cell migration, clonogenic survival and radioresistance in the AKT1 knockout MDA-MB-231BR cell lines, especially in the AKT1_KO_C1 clone, we took a deeper look into the WGS-data and additionally performed gene expression analysis via RNAseq.

After excluding variants (SNPs and Indels), also present in the MDA-MB-231BR_Ctrl cells, we identified 118 (corresponding to 83 genes) and 123 (corresponding to 83 genes) genomic sequence variants that were unique to the AKT1_KO_C1- and AKT1_KO_G4-clone, respectively ( Supplementary Table 2 ). Missense mutations were the most common mutation type in both AKT1 clones, followed by the splice site mutation, frame shift and in frame variants. Among them, those showing a deregulation at the mRNA level (RNAseq-data) and therefore with a possibly higher impact on impairing protein function were closely evaluated ( Supplementary Table 2 , bold: AKT1, GNL2, DOCK7, FLG, PLD2, ZNF236, FEM1A, LRRC75B, MAST4, EIF4G3, NBPF10, PLXNA2, MUC5B, PUS3, VPS35L, SF3B3, RACK1, HLA-A, HLA-C) regarding its potential impact on cancer cell stemness or resistance to ionizing radiation. Here, except for AKT1, no substantial association between the identified variant-associated factors and a more aggressive and stem-like phenotype or an increased sensitivity to radiation therapy in BC could be found.

For the transcriptome comparison between MDA-MB-231BR_Ctrl, AKT1_KO_C1 and AKT1_KO_G4, DEG were mapped to cellular pathways from the Gene Ontology database. Depicted in Figure 5 , the enrichment ratio is presented for both AKT1_KO cell lines in comparison to Ctrl ( Figure 5A ). 14 deregulated pathways in AKT1_KO_C1 and 13 deregulated pathways in AKT1_KO_G4 compared to Ctrl were found. Among them, 10 pathways were commonly deregulated in both knockout AKT1 clones, including extracellular matrix, extracellular matrix organization, cell motility and locomotion.

To have a closer look at the molecular mechanism of the increased cell migration, clonogenic survival as well as radioresistance in the MDA-MB-231BR AKT1_KO cell lines, gene sets related to stem cell differentiation, stem cell proliferation, stem cell division and radiosensitization were appropriated for the detection of deregulated genes to explain our unexpected data, especially in AKT1_KO_C1. As represented in a volcano plot, in the AKT1_KO_C1, 36 deregulated genes were identified, 11 of which were up-regulated, and 25 genes were down-regulated ( Figure 5B ; Supplementary Table 3 ). In AKT1_KO_G4, we discovered 50 deregulated genes, 21 of which genes were up-regulated and 29 genes of which were down-regulated ( Figure 5C ; Supplementary Table 3 ). Among them, 11 genes were deregulated in both AKT1 knockout clones compared to Ctrl ( Supplementary Table 3 , bold). In brief, the 10 genes AKT1, CEBPA, CTSO, CYBB, GPR68, ID1, ID4, METTL15, PBX1 and PTGFRN were concordantly significantly deregulated in AKT1_KO_C1 and AKT1_KO_G4, whereas the gene AMOT was up-regulated in the clone C1 (log2Foldchange=2.436, p-value=2.569x10^-38) and down-regulated in the clone G4 (log2Foldchange=-3.179, p-value=4.126x10^-22). These data indicate a significantly altered gene expression profile in the two AKT knockout cell lines and could explain their aggressive and stem cell-like phenotype in comparison to MDA-MB-231BR_Ctrl cell line.