Disruption of STAT5A and NMI signaling axis leads to ISG20-driven metastatic mammary tumors

Abstract

Molecular dynamics of developmental processes are repurposed by cancer cells to support cancer initiation and progression. Disruption of the delicate balance between cellular differentiation and plasticity during mammary development leads to breast cancer initiation and metastatic progression. STAT5A is essential for differentiation of secretory mammary alveolar epithelium. Active STAT5A characterizes breast cancer patients for favorable prognosis. N-Myc and STAT Interactor protein (NMI) was initially discovered as a protein that interacts with various STATs; however, the relevance of these interactions to normal mammary development and cancer was not known. We observe that NMI protein is expressed in the mammary ductal epithelium at the onset of puberty and is induced in pregnancy. NMI protein is decreased in 70% of patient specimens with metastatic breast cancer compared to primary tumors. Here we present our finding that NMI and STAT5A cooperatively mediate normal mammary development. Loss of NMI in vivo caused a decrease in STAT5A activity in normal mammary epithelial as well as breast cancer cells. Analysis of STAT5A mammary specific controlled genetic program in the context of NMI knockout revealed ISG20 (interferon stimulated exonuclease gene 20, a protein involved in rRNA biogenesis) as an unfailing negatively regulated target. Role of ISG20 has never been described in metastatic process of mammary tumors. We observed that overexpression of ISG20 is increased in metastases compared to matched primary breast tumor tissues. Our observations reveal that NMI-STAT5A mediated signaling keeps a check on ISG20 expression via miR-17-92 cluster. We show that uncontrolled ISG20 expression drives tumor progression and metastasis.

Fig. 1. NMI and STAT5A show concurrent expression pattern during mammary gland differentiation.

a NMI and STAT5A expression within computational reconstruction of the luminal compartment generated from single-cell RNA-seq mouse mammary cells different stages of differentiation. b t-distributed stochastic neighbor embedding (t-SNE) overlaid NMI and STAT5A expression in different Luminal and Basal clusters. c Immunohistochemical staining for STAT5A and NMI on mouse mammary tissues during different stages of development (puberty, mature non parous, pregnancy d10.5, pregnancy d14.5, lactation d1, lactation d4, involution d1, involution d3, and involution d8). d IHC immunoscore for STAT5A and NMI staining on mouse mammary tissues during differentiation. Expression was the highest during pregnancy and lactation, then subsided during the course of involution. The pattern of expression showed a significant positive correlation between NMI and STAT5A.

Fig. 2. NMI and STAT5A show concurrent expression pattern and nuclear localization in HC11 mammary cell differentiation.

a Western blot for β-Casein, NMI, Phospho-STAT5A, and STAT5A in undifferentiated HC11 cells and during the induction of differentiation using Dexamethasone, Insulin and Prolactin at hours 24, 48, and 72 h from the start of induction. b Normalized integrated density values for western blot signals for β-casein, NMI, P-STAT5A, and STAT5A in HC11 during the course of differentiation. c Fold expression changes of NMI and STAT5A using RTq-PCR in undifferentiated HC11 cells and during the induction of differentiation using Dexamethasone, Insulin, and Prolactin at hours 24, 48, and 72 from the start of induction. d Undifferentiated HC11 cells that were embedded in Matrigel and stimulated by DIP to differentiate into organized mammary acini. Right panel represents bright field images and left panel represents fluorescence imaging of E-cadherin in green to represent cell-to-cell junction and DAPI staining for nucleus in blue. e Immunofluorescence staining for NMI (Green) and STAT5A (red) in undifferentiated and differentiated HC11. DAPI was used to stain the nucleus, magnified pictures represent nuclear translocation of NMI and STAT5A in differentiated cells. f Graph represents mean nuclear florescence intensity quantification of NMI and STAT5A in undifferentiated and differentiated cells. This quantitation was conducted for all cells in the Z-stack of 3D organoid. g Graph represents Nmi and STAT5A colocalization correlation in nuclei of undifferentiated (n = 20) and differentiated (n = 15) cells.

Fig. 3. Silencing NMI hindered JAK/STAT signaling mediated mammary differentiation.

a GSEA of NMI wild-type compared to NMI knockout mammary mouse tissues at day 1 of lactation in comparison to list of genes upregulated after knockdown of JAK2 in HEL cells (left panel) and in comparison, to genes affected in lactating mouse mammary tissues after deletion of STAT5 enhancers (Right panel). b Immunohistochemical staining for STAT5A on mouse mammary tissues at lactation d1 in NMI fl/fl (n = 5) vs NMI −/− (n = 6) mice. Graph represents quantified nuclear STAT5A staining intensity, which is significantly lower in NMI−/− mice tissues (p = 0.017). c Western blot of Nmi, P-STAT5A, STAT5A in nontransfected control vs shNMI HC11 cells. d Undifferentiated shNMI HC11 cells that were embedded in Matrigel and stimulated by DIP to differentiate shNMI cells failed to differentiate into organized mammary acini. e Immunofluorescent staining for NMI (Green) and STAT5A (red) in control and shNMI HC11 cells that were embedded in Matrigel and stimulated to differentiate with DIP. DAPI was used to stain the nucleus. Pictures show loss of NMI and STAT5A nuclear translocation after knocking down NMI. f Luciferase assay for measuring activity of β-casein promotor. Data show decreased activity in T47D shNMI cells (p = 0.08). g STAT5 response element (STAT5-RE) activity was measured by a STAT5 response element reporter luciferase assay with and without prolactin. Data show diminished activity in T47D shNMI cells after prolactin stimulation.

Fig. 4. NMI/STAT5A axis is downregulated in breast cancer and its expression is distinctive for less frequent metastasis and good prognosis.

a Correlation of STAT5A and NMI immunoscore in human breast cancer TMA of normal breast tissues (n = 12), primary breast tumors (n = 48), and metastatic LN (n = 36), graph represents a significant moderate correlation between NMI and STAT5A score (r = 0.39) (p ≤ 0.0001). b Immunohistochemical staining for STAT5A and NMI on matched normal breast vs primary breast tumor (left panel) and matched primary breast tumor vs metastatic LN (right panel) c Immunoscore for STAT5A and NMI from matched human normal breast vs primary breast tumor and matched primary breast tumor vs metastatic LN. (Normal vs tumor NMI (P = 0.17) STAT5A (p = 0.001)) (Primary vs metastatic NMI (p = 0.05) STAT5A (p = 0.004)). The number in the circle represents immunoreactive score. The shade of color for individual circle is assigned it to distinguish matched sample. The line connects the matched samples from the same patients. d Heatmap of NMI, STAT5A, and JAK2 RNA-seq expression in TCGA 1101 primary human breast cancer. e Graph represents significant positive correlation between expression of STAT5A and NMI in TCGA (r = 0.39) (p ≤ 0.0001). f TCGA data-based KM survival curve for STAT5A RNA-seq low and high-expressing patients. Graph shows significantly higher overall survival (OS) (p = 0.002) and Progression-free interval (PFI) (p = 0.047) in STAT5A high-expressing patients.

Fig. 5. NMI/STAT5A axis downregulate ISG20 through miRNA hsa-miR-20.

a Heatmap of RNA-seq expression of STAT5 enhancer related genes in mammary mouse tissues for wild-type (n = 2) vs NMI KO (n = 3) mice mammary tissue. b Heatmap of fold change RNA-seq expression of Stat5 enhancer related genes in T47D shNMI vs control and 231 NMI vs control. c Fold expression changes of ISG20 using RTq-PCR in T47D vector control and shNMI, MDA-MB-231 scrambled control and NMI, ISG20 expression was significantly higher in T47D shNMI (p = 0.03) and lower in MDA-MB-231 NMI (p ≤ 0.0001). d WB of ISG20 and NMI proteins in T47D control and shNMI and MDA-MB-231 control and NMI expressing cells. e Schematic illustration of ISG20 gene structure with predicted binding sites of miR-17–92 cluster at the 3’UTR region. f Heatmap representing fold changes of miR-17–92 cluster expression in miRNA array for MCF10CA1cl.D NMI, MDA-MB-231 NMI, MCF10A shNMI, and T47D shNMI compared to controls, scale represents log2 fold changes. g Western blot of ISG20 in T47D cells 48 h after being supplemented with miRNA mimics (100 nM) Hsa-miR-17-3p, Hsa-miR-17-5p, Hsa-miR-20a-3p, or Hsa-miR-20a-5p. Graph represents IDV quantification fold changes.

Fig. 6. ISG20 negatively correlates with STAT5A and its expression promotes aggressive metastatic phenotype.

a Photomicrographs showing immunohistochemical staining for ISG20 using matched normal breast compared with corresponding primary breast tumor (left panel) and matched primary breast tumor compared with corresponding metastatic LN (right panel). b Immunoscore of ISG20 staining comparing matched human normal breast vs primary breast tumors and matched primary breast tumors vs corresponding metastatic LN. (Average IRS score 6.0 in normal vs 7.2 in tumor (p = 0.15)) (7.7 in Primary vs 9.5 in metastatic (p = 0.0002)). c Correlation between STAT5A and ISG20 immunoscores in human breast cancer TMA consisting of normal breast tissues (n = 12) primary breast tumors (n = 48) and metastatic LN (n = 36), graph represents a significant negative correlation between ISG20 and STAT5A score (r = 0.27) (p = 0.008). d ISG20 mRNA expression from TCGA data (RNA-seq) for breast cancer- normal breast tissues and breast tumor tissues are compared. ISG20 expression is significantly higher in breast cancer compared to normal breast tissue (p ≤ 0.0001). e Representative images of transmembrane migration for control or MDA-MB-468 and MDA-MB-231 cells overexpressing ISG20. Graphs represent significantly higher migrated cells/field in ISG20 expressing MDA-MB-468 (p = 0.0003) and MDA-MB-231 (p ≤ 0.0001). f Representative images for control or ISG20 MDA-MB-468 and MDA-MB-231 cells Invading through Matrigel invasion chamber. Graphs represent significantly higher invaded cells/field in ISG20 MDA-MB-468 (p ≤ 0.0001) and MDA-MB-231 (p ≤ 0.0001). g Images from fluorescence labelled MDA-MB-231 control or ISG20 overexpressing cells in PUMA assay. Nude mice were injected in the lateral tail (1 × 10⁵). Agar perfused lung slices were cultured on Surgifoam for 32 days. Metastatic burden was quantified as corrected total cell fluorescence (CTCF). MDA-MB-231 cells overexpressing ISG20 show a higher metastatic burden than the control cells (p ≤ 0.0001).

Fig. 7. Schematic summary of findings.

In normal breast NMI and STAT5A signaling keeps ISG20 expression in check, through expression of miR-20a and cells undergo normal differentiation. In cancer, NMI and Stat5A expression is compromised that reduces the check on ISG20 expression. Increased ISG20 promotes invasive progression that leads to metastatic colonization at a secondary site.