Let-7 microRNA controls invasion-promoting lysosomal changes via the oncogenic transcription factor myeloid zinc finger-1

Abstract

Cancer cells utilize lysosomes for invasion and metastasis. Myeloid Zinc Finger1 (MZF1) is an ErbB2-responsive transcription factor that promotes invasion of breast cancer cells via upregulation of lysosomal cathepsins B and L. Here we identify let-7 microRNA, a well-known tumor suppressor in breast cancer, as a direct negative regulator of MZF1. Analysis of primary breast cancer tissues reveals a gradual upregulation of MZF1 from normal breast epithelium to invasive ductal carcinoma and a negative correlation between several let-7 family members and MZF1 mRNA, suggesting that the inverse regulatory relationship between let-7 and MZF1 may play a role in the development of invasive breast cancer. Furthermore, we show that MZF1 regulates lysosome trafficking in ErbB2-positive breast cancer cells. In line with this, MZF1 depletion or let-7 expression inhibits invasion-promoting anterograde trafficking of lysosomes and invasion of ErbB2-expressing MCF7 spheres. The results presented here link MZF1 and let-7 to lysosomal processes in ErbB2-positive breast cancer cells that in non-cancerous cells have primarily been connected to the transcription factor EB. Identifying MZF1 and let-7 as regulators of lysosome distribution in invasive breast cancer cells, uncouples cancer-associated, invasion-promoting lysosomal alterations from normal lysosomal functions and thus opens up new possibilities for the therapeutic targeting of cancer lysosomes.

Fig. 1. MZF1 expression is increased in primary breast cancer samples and its expression induces cancer cell invasion in poorly invasive breast cancer cells.

a Examples of IHC staining of MZF1 (brown) in normal breast (upper panel), grade 1 breast cancer tissue (DCIS; middle panel), grade 1–2 breast cancer tissue (lower panel). Sections were counterstained with hematoxylin (blue) and images were taken with 20× magnification. b Association of MZF1 protein expression with tumor grade in primary breast cancer samples in TMAs including 321 primary samples of normal and breast cancer tissue. TMAs were analyzed by ACIS-assisted quantitative IHC and ACIS scores of duplicate cores were averaged and plotted against the breast cancer grade. The statistical difference in MZF1 expression between Normal and Grade 1–2 (IDC) is indicated by a * implicating a p-value < 0.05 according to a Mann–Whitney test. Mean values +/− SD is marked in red. c MZF1 expression in normal, benign and primary breast tumors of grade 1–3 from the same TMAs as in (b). d Invasion of MCF7 cells stably transfected with a doxycycline-inducible expression vector containing MZF1 (MCF7-MZF1 1–6) or an empty vector (MCF7-vector) in 3D Matrigel in the absence (−Dox) or presence of doxycyline (+Dox). Images were taken with 10x magnification. Quantification of the average extend of outgrowth in (d) was done using ImageJ. For each of 4 spheres per treatment, the 10 greatest distances traveled by invading cells were estimated and a mean outgrowth was calculated for each treatment. Data presented is an average of 2 independent experiments and as % of non-induced control (−Dox) and the comparison of −Dox and +Dox in MCF7-MZF1 1–6 cells is assigned with **** indicating p < 0.0001 in a Welch’s t-test. e Immunoblot analysis of MZF1 expression in MCF7-vector and MCF7-MZF1 1-6 (+Dox) and without (−Dox) doxycycline. β-actin was used as a control of equal loading. The immunoblot from a single control experiment was quantified using ImageJ, normalized to β-actin and presented as % of MCF7-vector −Dox

Fig. 2. The MZF1 3′-UTR contains a putative let-7 target site and MZF1 expression correlates negatively with several let-7 family members and positively with LIN28A in primary breast cancer samples.

a Alignment of 9 let-7 family members to their putative target site in the MZF1 3′-UTR (red). All let-7 family members are predicted to bind to the target site via a 7mer seed region (green), with the exception of let-7e binding with an 8mer seed region. b Correlation plots for the correlation between let-7 and MZF1 RNA expression in primary breast cancer samples included in TCGA data. Pairs that show positive correlation are depicted with red circles. Pairs that show negative correlation are depicted with blue circles. The Spearman correlation coefficient is depicted using the radius of the circle: the larger the coefficient, the larger the radius. c Correlation plots of MZF1 and LIN28A expression in TCGA data (all stages). Highest 6% expressing LIN28A samples have been omitted from the figure to better show the correlation. The correlation is assigned with *** indicating p < 0.001. d Correlation plots of MZF1 and LIN28B expression in TCGA (all stages). e Association between MZF1 and Lin28A protein expression quantified from the same TMAs as described in Fig. 1a–c. The Spearman correlation is significant with p < 0.0001

Fig. 3. Let-7 miRNAs regulate MZF1 expression by directly binding to its 3′-UTR.

a Representative IHC staining of MZF1 in sections made from formalin-fixed, paraffin-embedded (FFPE) MCF10A and MCF7 cell pellets. Sections were counterstained with hematoxylin (blue) and images were taken with 40x magnification. MZF1 expression in MCF10A and MCF7 cells was quantified using ACIS-assisted analysis. b Let-7 expression in MCF10A and MCF7 cells. Expression of let-7d, let-7e, let-7f, let-7g, and let-7i was quantified using TaqMan miRNA assay and qPCR. The relative miRNA quantity was calculated by 2-ΔCt using U6 RNA as an internal reference. U6 RNA was expressed at similar levels in MCF10A and MCF7 cells. Data are presented as the means of two independent experiments with three technical replicates ± SD. The comparison of MCF10A and MCF7 cells according to let-7e- and let-7g-expression is assigned with ** indicating p < 0.01 and **** indicating p < 0.0001 in Welch’s t-test. c The effect of let-7d, let-7e and let-7 g mimics (20 nM) on MZF1 mRNA level relative to non-targeting control mimic in MCF7 cells. MZF1 mRNA expression was quantified by RT-qPCR with PPIB as reference gene. Data are presented as the mean of six (let-7d and let-7e) and five (let-7g) independent experiments ± SD and significance is indicated with * for p < 0.05 in a Welch’s t-test. d The effect of miRNA inhibitors LNA1, LNA2 and LNA3 (20 nM) targeting different let-7 family members on MZF1 mRNA expression relative to an LNA negative control in MCF7 cells. Data are presented as the mean of three independent experiments ± SD and significance is indicated with * and ** for p < 0.05 and p < 0.01 in a Welch’s t-test, respectively. e Immunoblot-analysis of MZF1 protein expression in response to a let-7 pool (let-7d, let-7e, and let-7g, 20 nM) or LNA pool (LNA1 and LNA3, 20 nM). β-actin was used to control the equal loading. The immunoblot is a representative of three independent blots that were quantified using ImageJ. Data are presented as the mean of three experiments ± SD relative to control. Significance is indicated with * for p < 0.05 in a Welch’s t-test. f Immunoblot-analysis of MZF1 protein expression in response to a transfection with 20 nM of let-7d, let-7e, and let-7 g mimic and a non-targeting control in SKBR3, MDA-MB-231 and MDA-MB-436 cells. β-actin was used to control the equal loading. The immunoblot is a representative of three independent blots. g Luciferase reporter assay measuring the hluc + − and hRluc-response of psiCHECK-2-MZF1, Mutated psiCHECK-2-MZF1 and a positive control psiCHECK-2–2xlet-7 to mimics of let-7d, let-7e and let-7g (20 nM). Data are presented as the mean of 12 independent experiments for psiCHECK-2-MZF1 and mutated psiCHECK-2-MZF1 and 4 independent experiments for psiCHECK-2–2xlet7. Data are presented as the mean of experiments ± SD relative to control mimic and significance is indicated with *, **, ***, and **** for p < 0.05, 0.01, 0.001, and 0.0001 in a Welch’s t-test, respectively

Fig. 4. Let-7 miRNAs inhibit invasion of invasive breast cancer cells by reducing MZF1 expression.

a Invasion of MCF7-p95ΔNErbB2 cells in 3D Matrigel upon transfection with 20 nM control, let-7d, let-7e, and let-7g mimic and MZF1 siRNA. Images of the invasive spheres were taken at 10× magnification. Quantification of the average extend of outgrowth in (a) was done using ImageJ. For each of 5 spheres per treatment, the 10 greatest distances traveled by invading cells were estimated and a mean outgrowth was calculated for each treatment. Data are presented as % of control mimic and the effect of let-7e and MZF1kd is assigned with *, **, and *** indicating p < 0.05, p < 0.01, and p < 0.001, respectively, in a Welch’s t-test. b Invasion of MCF7-p95ΔNErbB2 cells in 3D Matrigel upon transfection with a control and 20 nM let-7e mimic in the absence and presence of transiently expressed WT MZF1 lacking the 3′-UTR. Images of the invasive spheres were taken with 10× magnification. Quantification is done as described in a. Data are presented as % of control and the effect of let-7e is assigned with * indicating p < 0.05 in a Welch’s t-test. c Immunoblot analysis of MZF1 protein expression in MCF7-p95ΔNErbB2 cells transfected with 20 nM of control and let-7e mimic in the absence and presence of transient overexpression of WT MZF1. β-actin was used to control for equal loading. The immunoblot was quantified using Image Studio Lite. Data are presented as relative to control mimic treatment in the absence of transient WT MZF1 overexpression. d Immunoblot analysis of ErbB2 protein expression in MCF7-p95ΔNErbB2 cells response to forced let-7e expression (20 nM). β-actin was used to control the equal loading. The immunoblot is a representative of three independent experiments that were quantified using Image Studio Lite. Data are presented as the mean of experiments ± SD relative to control. The effect was non-significant according to a Welch’s t-test

Fig. 5. Let-7 miRNAs reverse ErbB2-induced peripheral distribution of lysosomes by reducing MZF1 expression.

a Confocal immunofluorescence microscopy images of the lysosomal distribution in MCF7-p95ΔNErbB2 cells treated with DMSO or 5 μM lapatinib for 24 h. After treatment, cells were fixed and stained for the lysosomal membrane protein LAMP2 (green), cytoskeletal α-tubulin (red) and nucleus with Hoechst (blue). Quantification of (a) was done by characterizing the lysosomal distribution as predominantly perinuclear, scattered or peripheral in 8–15 cells per 63× image in five images per treatment. Data represents the mean percentage distribution of five images ± SD. The effect of lapatinib on the peripheral pool of lysosomes is assigned with *** indicating p < 0.001 in a Welch’s t-test. b Confocal immunofluorescence microscopy images of the lysosomal distribution in MCF7-p95ΔNErbB2 cells transfected with 20 nM of a non-targeting control, let-7d, let-7e, and let-7g mimic and MZF1 siRNA (MZF1kd) and the non-targeting control and let-7e mimic in the presence of transiently expressed WT MZF1 lacking the 3′-UTR. Quantification of (b) was done as described in for (a). The effect of let-7 mimics and MZF1 siRNA on the peripheral pool of lysosomes is assigned with *, **, and *** indicating p < 0.05, p < 0.01, and p < 0.001 in a Welch’s t-test. The data presented in A and B are representative of three independent experiments