Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Mar 13;9(1):1044.
doi: 10.1038/s41467-018-03283-z.

Integrated genomics and functional validation identifies malignant cell specific dependencies in triple negative breast cancer

Affiliations

Integrated genomics and functional validation identifies malignant cell specific dependencies in triple negative breast cancer

Nirmesh Patel et al. Nat Commun. .

Abstract

Triple negative breast cancers (TNBCs) lack recurrent targetable driver mutations but demonstrate frequent copy number aberrations (CNAs). Here, we describe an integrative genomic and RNAi-based approach that identifies and validates gene addictions in TNBCs. CNAs and gene expression alterations are integrated and genes scored for pre-specified target features revealing 130 candidate genes. We test functional dependence on each of these genes using RNAi in breast cancer and non-malignant cells, validating malignant cell selective dependence upon 37 of 130 genes. Further analysis reveals a cluster of 13 TNBC addiction genes frequently co-upregulated that includes genes regulating cell cycle checkpoints, DNA damage response, and malignant cell selective mitotic genes. We validate the mechanism of addiction to a potential drug target: the mitotic kinesin family member C1 (KIFC1/HSET), essential for successful bipolar division of centrosome-amplified malignant cells and develop a potential selection biomarker to identify patients with tumors exhibiting centrosome amplification.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1
Fig. 1
Integrative gene addiction identification and validation. Schematic representation of bioinformatics based “Target ID” and functional validation RNAi experiment. a Composition of the Guy’s TNBC-enriched cohort of breast cancers, used as a source of DNA and RNA for this study. Two complementary approaches for candidate gene addiction identification. In green, copy number-dependent gene expression analysis shows initial filter for copy number gain/amplification and correlation to gene expression and subsequent Target ID algorithm. This pipeline consisted of a weighted scoring system for all genes based on copy number (CN), gene expression (GEX), gene, and clinical annotation listed here and in the Methods section. In purple, gene expression-centered analysis followed by manual curation based on gene annotation and literature evidence of an involvement in malignancy. b Workflow and hit selection criteria from primary functional validation RNAi experiment. *Top 10 genes were subject to different criteria as outlined in Results and Methods. c Example functional validation for TTK, using pool of four siRNAs across panel of cell lines. Mean normalized percent inhibition (NPI) from three independent experiments is plotted and error bars represent the standard error of the mean (SEM), n = 3. d Example of data from primary functional validation carried out in CAL51 cells. Data points represent the mean NPI from three independent replicates. Dashed line represents cut-off for positive hits at 18.01% NPI. e Mean NPI from CAL51 cell line after deconvolution of pool of TTK siRNAs used independently. Error bars represent SEM, n = 3. f mRNA knockdown of individual siRNAs in CAL51 cells. Knockdown was evaluated by qPCR and represented as mean percentage of knockdown compared to non-silencing control. Error bars represent SEM, n = 3
Fig. 2
Fig. 2
A subset of tumor addiction genes that are co-upregulated have roles in cell cycle progression, mitosis and DNA damage response. Copy number (a) and gene expression (b) levels of 37 tumor addiction genes were pairwise correlated and tested for statistical significance using Pearson method in the TNBC tumors of the Guy’s TNBC-enriched cohort (n = 82). Their correlation coefficients were hierarchically clustered using Ward distance method, and displayed as levelplots. In the gene expression correlation heatmap highly correlated genes are surrounded by black rectangles, representative of underlying clusters from hierarchical clustering method whereas a cross denotes insignificant p-value as per level of 0.05. c Interconnectivity among the 13 co-upregulated tumor addiction genes is displayed based on STRING networks. Genes with shared functional roles are outlined
Fig. 3
Fig. 3
KIFC1 is a validated tumor addiction gene that is upregulated in TNBCs. a Primary pooled siRNA oligo validation data for KIFC1. Mean NPI are plotted and error bars represent the SEM, n = 3. b Mean NPI from three independent of HCC1143 cell line after deconvolution of pool of KIFC1 siRNAs used independently. Error bars represent SEM, n = 3. c mRNA knockdown of individual siRNAs in HCC1143 cells. Knockdown was evaluated by qPCR and represented as percentage of knockdown compared to non-silencing control from three technical replicates. d Copy number alterations of KIFC1 correlated with its gene expression across all breast cancers in the Guy’s TNBC-enriched cohort of breast cancers, (e) TCGA BRCA and (f) METABRIC data sets. g KIFC1 gene expression across the PAM50 breast cancer subtypes in the Guy’s TNBC-enriched cohort of breast cancers, (h) TCGA BRCA and (i) METABRIC data sets. Box-and-whisker plots showing median center line, 25% and 75% box limits and range of expression, non-paired two-sided Wilcoxon rank sum test; *p < 0.05, ****p < 0.0001
Fig. 4
Fig. 4
KIFC1 is specifically required for survival of cancer cells exhibiting centrosome amplification. a Centrosome amplification (CA) scores for panel of breast cancer cell lines. Cell lines were dichotomized into low CA (black) and high-CA (red) groups. b Representative immunofluorescence images showing centrosome marker Aurora A (green) and centriole marker CP110 (red) used for calculating CA score. Scale bar is 5 μm. Top right inset shows red channel, bottom right inset for each image shows green channel. c Mean normalized percent inhibition (NPI) of panel of cell lines dichotomized into low and high-CA groups for siRNA #2, #4, #5, and #6. Error bars represent the SEM, n > 3. Student's t test: **p < 0.01, ***p < 0.001, ****p < 0.0001. d Colony formation assay of two low centrosome-amplified cell line (CAMA1 and SUM149) and two high-centrosome-amplified cell lines (MDA-MB-231, HCC1954 and BT20) infected with either non-targeting shRNA (NT) or shRNA-targeting KIFC1 (KIFC1 shRNA). Cells were grown for 14–21 days in the presence or absence of doxycycline, fixed and stained in crystal violet and colonies quantified. Mean surviving fraction normalized to NT, error bars represent the SEM, n = 3. One-way ANOVA with Tukey’s multiple comparisons test: ***p < 0.001, ****p < 0.0001. e NPI of MDA-MB-231 cells infected with either GFP alone (GFP) or GFP-tagged RNAi-resistant KIFC1 (GFP-KIFC1r) with KIFC1 siRNA #4 and #6. Error bars represent the SEM, n = 3, one-way ANOVA with Tukey’s multiple comparisons test: *p < 0.05, **p < 0.01. f Colony formation assay of MDA-MB-231 cells infected with inducible KIFC1 shRNA and infected with inducible RNAi-resistant HA-tagged KIFC1 (KIFC1-HAr) or with empty vector (EV) control. Mean surviving fraction normalized to NT, error bars represent SEM, n = 3. One-way ANOVA with Tukey’s multiple comparisons test: ****p < 0.0001. g Nude hosts were orthotopically injected with either MDA-MB-231 or HCC1954 cells with inducible KIFC1 shRNA and were treated with (red) or without (black) doxycycline when tumors reached >2 × 2 mm (4.2 mm3). Mean tumor volumes at time points indicated, error bars represent the SEM, from two independent experiments. Two-way ANOVA with Sidak’s multiple comparisons test: *p < 0.05, **p < 0.01, ****p < 0.0001
Fig. 5
Fig. 5
Cell lines with centrosome amplification undergo multipolar mitoses followed by mitotic catastrophe in the absence of KIFC1. a Representative immunofluorescence images of mitotic spindles in low-CA cell line (CAL51) and high-CA cell lines (HCC1143 and SKBR3) with control or KIFC1 siRNA or mock transfection. Spindles stained with Aurora A (green) and/or Eg5 (red). Scale bar is 5μm. b Mean percentage of multipolar mitotic cells. Error bars represent the SEM, n = 3. c Images of time-lapse of MDA-MB-231 cells expressing NT shRNA showing a bipolar mitosis. d Images of time-lapse of MDA-MB-231 expressing KIFC1 shRNA showing prolonged mitosis and subsequent apoptosis. e. Images of MDA-MB-231 cells with inducible KIFC1 shRNA showing an abnormal multipolar mitosis. Scale bar is 5 μm
Fig. 6
Fig. 6
Therapeutic induction of centrosome amplification in tumors increases the efficacy of KIFC1 silencing. a MDA-MB-231 cells were treated with five serial concentrations of cisplatin and DMSO control and scored for centrosome amplification (CA). Table shows the level of CA at each concentration of cisplatin. b Surviving fraction of MDA-MB-231 cells with inducible NT shRNA with (red) or without (black) doxycycline. Data points represent the mean, error bars represent the SEM, n = 3. Extra sum-of-square F-test (F-value 0.41). Right, representative images of colony formation wells for each treatment condition. c Surviving fraction of MDA-MB-231 cells with inducible KIFC1 shRNA with (red) or without (black) doxycycline. Data points represent the mean, error bars represent the SEM, n = 3. Extra sum-of-squares F-test (F-value 57.59). Right, representative images of colony formation wells for each treatment condition. Note, five times more MDA-MB-231 cells with KIFC1 shRNA (Dox) were plated due to low colony forming ability of cells without KIFC1. d Nude hosts were injected with MDA-MB-231 cells with inducible KIFC1 shRNA and were treated with or without doxycycline and either vehicle or 5 mg kg−1 of cisplatin when tumors reached >2 × 2 mm or 4.2 mm3. Mean tumor volume at time points indicated, error bars represent the SEM, from two independent experiments. Two-way ANOVA with Tukey’s multiple comparison test: **p < 0.01, ****p < 0.0001 (only shown for final time point). e Nude hosts were injected with MDA-MB-231 cells with inducible KIFC1 shRNA and were treated with either vehicle or 5 mg kg−1 of cisplatin when tumors reached >33 mm3. Tumors taken 4 days after start of treatment in the no dox group were stained with pericentrin by IHC and the percentage of abnormal centrosomes was scored for vehicle and cisplatin treatments arms. Mean percentage of cells with centrosome abnormalities, error bars represent the SEM, n = 4. Student's t test: *p < 0.05
Fig. 7
Fig. 7
Pericentrin abnormality score: a potential predictive biomarker for sensitivity to KIFC1 inhibition. a Normal breast tissue section stained for pericentrin as a centrosome marker and below, a scatter graph of DAB staining area versus DAB mean intensity. Cut-off for normal centrosome size was set at 7 µm2. b Tumor 1 with centrosomes that appear normal. Below, scatter graph of DAB staining area vs DAB mean intensity showing a PCAB score of 1.19%. c Tumor 2 with centrosomes that appear abnormal. Below, scatter graph of DAB staining area vs DAB mean intensity showing a PCAB score of 12.34%. d PCAB score vs NPI (%) upon KIFC1 silencing across panel of breast cancer cell lines. Linear regression analysis, r2 = 0.71, p < 0.05. Scale bars represent 25 μm. e. PCAB score of TMAs of cohort of 82 TNBCs. Dashed red line depicts the median PCAB score (32%) of breast cancer cell lines sensitive to KIFC1. f Kaplan–Meier curves illustrating the duration of recurrence free survival according to a 20% PCAB cut-off. Wald test, p < 0.05, hazard ratio (HR) = 1.95, confidence intervals (CI) = 1–3.89

References

    1. Reis-Filho JS, Tutt ANJ. Triple negative tumours: a critical review. Histopathology. 2008;52:108–118. doi: 10.1111/j.1365-2559.2007.02889.x. - DOI - PubMed
    1. Lehmann BD, et al. Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J. Clin. Invest. 2011;121:2750–2767. doi: 10.1172/JCI45014. - DOI - PMC - PubMed
    1. Baselga J, Swain SM. Novel anticancer targets: revisiting ERBB2 and discovering ERBB3. Nat. Rev. Cancer. 2009;9:463–475. doi: 10.1038/nrc2656. - DOI - PubMed
    1. Hu Z, et al. Genome co-amplification upregulates a mitotic gene network activity that predicts outcome and response to mitotic protein inhibitors in breast cancer. Breast Cancer Res. 2016;18:70. doi: 10.1186/s13058-016-0728-y. - DOI - PMC - PubMed
    1. Zaman N, et al. Signaling network assessment of mutations and copy number variations predict breast cancer subtype-specific drug targets. Cell Rep. 2013;5:216–223. doi: 10.1016/j.celrep.2013.08.028. - DOI - PubMed

Publication types