Fig 2: Expression of the IRF1 and DTX3L proteins/mRNAs in breast cancer cell-lines and tissues. (A) IRF1 (Interferon-Responsive-Factor-1) and DTX3L (Deltex-E3-Ubiquitin-Ligase-3L) mRNAs were measured by PCR (Polymerase Chain Reaction) in the indicated 41 cell-lines (column graphs). Each value is the mean ± standard deviation (SD) of 3 replicate cultures and the results are normalized for the 18S mRNA. The Western blots illustrate the levels of IRF1 and DTX3L proteins. The 16 cell-lines belonging to our experimental panel are indicated in blue (basal cell-lines) and red (luminal cell-lines). The basal or luminal phenotype of all the other cell-lines is indicated by blue and red dots, respectively. Each well contains an equivalent amount of protein. CAL-851 and BT474 cells (internal loading controls) are marked in green. The asterisk underneath the indicated Western blots marks the cell-lines for which bands are not detectable. The positivity of each cell-line to HER2 (HER2+) or ER (ER+) as well as the negativity of each cell line to HER2, ER and PR (progesterone receptor) is indicated by the Triple-neg (Triple-negative) designation. For the whole Western blot figures, please see Supplementary Information. (B) and (C) The IRF1 as well as DTX3L mRNA and protein levels are compared following clustering of the cell-lines according to the luminal (Lum) and basal (Bas) phenotype or the presence/absence of the ER (Estrogen Receptor) and HER2 proteins as defined in (A). Each point represents a single cell-line. The quantitative protein data were obtained from the densitometric analysis of the Western blots shown in (A), following normalization with the internal loading controls. * Significantly different following the Student’s t-test, corrected for multiple testing (Benjamini–Hochberg) (p < 0.05). ** (p < 0.01). (D) and (E) The Box Plots show the IRF1 and DTX3L mRNA levels of breast cancer tissue samples determined from the RNA-seq data available in the TCGA database (1108 cases). The breast cancer cases are classified according to the PAM50 algorithm into 5 distinct groups: Luminal A = LumA, HER2-positive = HER2+, Luminal B = LumB, Normal-like = Nor-like, Basal = Bas. Each point represents an individual case. * Significantly different following pairwise t-test comparison adjusted for multiple testing according to the Benjamini–Hochberg correction method (p < 0.05). ** (p < 0.01).

Fig 3: Interaction networks of the genes up- and downregulated by ATRA in the retinoid-sensitive cell-lines. The 754 genes whose up- or downregulation is proportional to ATRA-sensitivity were used to construct an interaction network based on the encoded proteins (STRING database, https://string-db.org). The 2 upregulated modules (MHC Class-I Loading and presentation; Interferon Response) can be further condensed into a single group, as indicated by a red circle. The single genes belonging to this module are indicated by red dots. The higher the average upregulation afforded by ATRA, the larger the size of each red dot. The 4 downregulated modules (Cell-cycle, Histone-cluster I, Chromatin-packaging/organization, DNA-repair) can be further clustered into a single group marked by a blue circle. The single genes belonging to this module are indicated by blue dots. The lower the fold-change observed following ATRA treatment, the larger the size of each blue dot. The heat-maps of the ATRA/DMSO fold changes observed for the genes belonging to the up- (right) and downregulated (left) interaction networks in each cell-line are shown. The cell-lines are ranked from left to right according to their decreasing sensitivity to ATRA (decreasing ATRA-score value). Luminal cell-lines are indicated in red, while basal cell-lines are marked in blue. The IRF1 and DTX3L mRNAs are marked with a red circle.

Fig 4: Effects of ATRA and derived retinoids on IRF1 and DTX3L in breast cancer cell-lines and functional significance of the two proteins in ATRA-dependent growth inhibition. (A) IRF1 and (B) DTX3L protein and mRNA levels were determined in vehicle (DMSO)- or ATRA (1 μM)-treated cell-lines (24 h). IRF1 and DTX3L proteins were determined in the nuclear and cytosolic fraction by Western blot analysis. The column graphs above the Western blots show the densitometric results obtained. The column graphs underneath the Western blots show the effect exerted by ATRA on the expression levels of the IRF1 and DTX3L mRNAs (Taqman assays). (C) SK-BR-3 and (D) HCC-1599 cells were treated with vehicle and the indicated RAR agonist (1 μM), the γ-secretase inhibitors, DAPT [(2S)-N-[(3,5-Difluorophenyl)acetyl]-L-alanyl-2-phenyl]glycine-1,1-dimethylethyl-ester] (1 μM) and PF-3084014 [(2S)-2-[[(2S)-6,8-Difluoro-1,2,3,4-tetrahydro-2-naphthalenyl]amino]-N-[1-[2-[(2,2-dimethylpropyl)amino]-1,1-dimethylethyl]-1H-imidazol-4-yl]-pentanamide-dihydrobromide] (1 μM), PTX (paclitaxel, 0.01 μM), DOXO (doxorubicin, 0.2 μM) and VP16 (etoposide, 2.0 and 20 μM) for 24 h. Cellular extracts were subjected to Western blot analysis for IRF1, DTX3L and tubulin (TUB, loading control). IRF1/TUB = ratio of the densitometric results obtained for the IRF1 and Tubulin bands, DTX3L/TUB = ratio of the densitometric results obtained for the DTX3L and Tubulin bands. (E) SK-BR-3 cells were transfected with the small interfering RNAs, siIRF1(a), siIRF1(b), or the scrambled siNC control. 24 h following transfection, cells were treated with vehicle (DMSO) or the indicated concentrations of ATRA for another 96 h. Cell-growth was evaluated with sulforhodamine assays and the results are expressed as the % growth inhibition relative to the corresponding siNC control (lower panel). Values are the mean ± SD of 3 replicate cultures. * significantly different (Student’s t-test, p < 0.05; red asterisks: siIRF1(a) versus siNC, green asterisks: siIRF1(b) versus siNC). ** significantly different (Student’s t-test, p < 0.01; red asterisks: siIRF1(a) versus siNC, green asterisks: siIRF1(b) versus siNC). The effect of silencing on the IRF1 protein was evaluated after 24-h treatment with vehicle or ATRA (1 μM). Cell extracts were subjected to Western blot analysis for IRF1 and Tubulin (upper panel). IRF1/TUB = ratio of the densitometric results obtained for the IRF1 and Tubulin bands, DTX3L/TUB = ratio of the densitometric results obtained for the DTX3L and Tubulin bands. (F) SK-BR-3 cells were stably infected with shDTX3L(a), shDTX3L(b), the void lentiviral vector (shVOID) and a scrambled shRNA (shNC). Following antibiotic selection and isolation, the indicated cell populations were treated (24 h) with vehicle (DMSO) or ATRA (1 μM). Total cell extracts were subjected to Western blot analysis for DTX3L and TUB (upper panel). DTX3L/TUB = ratio of the densitometric results obtained for the DTX3L and Tubulin bands. Cell-growth (lower panel) was evaluated as in (E). Each value is the mean ± SD of 3 replicate cultures. SK-BR-3 = parental cell-line. * significantly different from shNC and shVOID (Student’s t-test, p < 0.05). ** significantly different from shNC and shVOID (Student’s t-test, p < 0.01). For the whole Western blot figures, please see Figure S19.

Fig 5: DTX3L interaction influences PARP14 ADP-ribosylation.. A) ADP-ribosylation assay using NAD+-biotin and 1 μg PARP14 constructs containing catalytic domain (CAT) or WWE or 14D/KHL domain (14DWC). 1 μg DTX3L D3-D4 was added before reaction was initiated with -NAD+ -biotin. Arrows indicate ADP-ribosylation products. B) 1 M Neutral hydroxylamine was incubated with the reaction products of the ADP-ribosylation assay to identify of modification of acidic residues for each PARP14 construct and in the presence of DTX3L KHL2-5 (expected position indicated by arrow). C) 190 nM PARP14 DWC truncation was included in an ADP-ribosylation assay in the presence of PARP9/DTX3L. Biotinylated ADP-ribosylation products (red) and the DTX3L species are detected with a specific antibody (green) such that ADP-ribosylated DTX3L appears yellow. D) Similarly, the upper band detected in the ADP-ribosylation assay was identified as PARP9 using a specific antibody (green).