Brain metastases (BM) result from the spread of primary tumors to the brain and are a leading cause of cancer mortality in adults. Secondary tissue colonization remains the main bottleneck in metastatic development, yet this “premetastatic” stage of the metastatic cascade, when primary tumor cells cross the blood–brain barrier and seed the brain before initiating a secondary tumor, remains poorly characterized. Current studies rely on specimens from fully developed macrometastases to identify therapeutic options in cancer treatment, overlooking the potentially more treatable “premetastatic” phase when colonizing cancer cells could be targeted before they initiate the secondary brain tumor. Here we use our established brain metastasis initiating cell (BMIC) models and gene expression analyses to characterize premetastasis in human lung-to-BM. Premetastatic BMIC engaged invasive and epithelial developmental mechanisms while simultaneously impeding proliferation and apoptosis. We identified the dopamine agonist apomorphine to be a potential premetastasis-targeting drug. In vivo treatment with apomorphine prevented BM formation, potentially by targeting premetastasis-associated genes KIF16B, SEPW1, and TESK2. Low expression of these genes was associated with poor survival of patients with lung adenocarcinoma. These results illuminate the cellular and molecular dynamics of premetastasis, which is subclinical and currently impossible to identify or interrogate in human patients with BM. These data present several novel therapeutic targets and associated pathways to prevent BM initiation.

Significance: These findings unveil molecular features of the premetastatic stage of lung-to-brain metastases and offer a potential therapeutic strategy to prevent brain metastases. Cancer Res; 78(17); 5124–34. ©2018 AACR.

Metastases to the brain (BM) are the most common neoplasm to affect the adult central nervous system, occurring in up to 40% of patients with cancer and at a rate 10 times greater than that of primary neural neoplasms (1). Survival of patients with BM is limited to mere weeks, extended to months upon administration of multimodal treatment (2). Despite the devastating clinical outcomes, the genetic and molecular events that govern metastatic development remain frustratingly difficult to predict. The process of metastasis is both complicated and extremely inefficient, where only a minute percentage of disseminated tumor cells are capable of surviving the lymphovascular system to establish metastatic tumors. Metastatic cells must first adapt to and seed this secondary environment, termed “premetastasis”; this tissue-colonization stage directly precedes formation of small micrometastases, and establishment of vasculature will promote larger macrometastatic growth. Understanding premetastasis is the largest barrier to metastatic development and tissue colonization, yet this stage remains poorly characterized in solid tumor-derived BM (3). Clinically, current diagnostic techniques require tumors to be of a certain size before they can be detected; theoretically, the delay between primary tumor formation and clinical diagnosis of metastatic growth, even with early tumor dissemination, provides a potential window for therapeutic intervention (4).

Significant investigation into the cancer genome has led to greater understanding of the evolving clonal architecture of tumors, exposing the coexistence of a dominant originating primary tumor clone along with multiple genetically distinct subclones that can give rise to recurrence and metastases (5, 6). Further lineage analyses have identified early and initiating conditions that define a “precancerous” stage in the progression of several primary cancers (7–9). Initiating events have similarly been explored for metastatic growth, identifying the conditional implementation of various mechanisms such as epithelial–mesenchymal transitions and angiogenesis by metastasis initiating cells (10). Unfortunately, there remains a dearth of knowledge of the mechanisms that promote “premetastatic” initiation and the tissue-colonization stage prior to establishment of tumor masses (11). Though many solid tumors undergo metastasis to the brain, the ability to recapitulate every intricate stage of this process in vivo is very difficult; as such current models are only able to mimic individual or partial stages at once. Additionally, the majority of current in vivo and clinical studies utilizes or analyzes established macrometastasis samples, failing to properly capture this temporally sensitive premetastatic stage. Systematic characterization of this premetastatic stage could provide more relevant avenues for therapeutic options in BM prevention as opposed to treating existing BM.

In the present work, we utilized our established patient-derived models of lung-to-brain metastasis to elucidate the molecular variances that underlie premetastatic initiation through focused study of human BMICs injected into immunocompromised mice via the intrathoracic route. Importantly, the premetastatic phase captures a stage of the metastatic cascade that can never be routinely biopsied or captured in humans, as metastatic cells seeding the brain without yet initiating a secondary tumor would represent subclinical disease that cannot be detected by either clinical symptoms or current surveillance neuroimaging techniques. We found these premetastatic BMICs (termed BMIT) to possess over 7,000 dysregulated genes, many of which are active in invasive but not in proliferative mechanisms; similar data have only recently been shown in C. elegans (12). Interestingly, these BMIT genes were also enriched in neural neoplasm and neurodegenerative pathways; studies have implicated an inverse correlation of genes involved in cancer development and neurodegenerative disorders, and where the gene expression profiles of our established lung and tumors and BM appear to support this, our premetastatic BMIT genes do not (13). Through Connectivity Map analysis (CMAP) of these BMIT genes and preliminary in vivo validation, we demonstrated that the dopamine agonist apomorphine inhibited BM development in vivo, presumably by inhibiting the premetastatic state. Further pharmacogenomic interrogation of the BMIT gene list identified 3 downregulated genes that are directly targeted by apomorphine, KIF16B, SEPW1, and TESK2, where administration of apomorphine restores expression. Lastly, interrogation of lung adenocarcinoma patient databases showed that decreased expression of these genes is associated with poor disease-free survival.

With this work we have successfully characterized a novel temporal genetic profile of premetastatic growth and have functionally validated the efficacy of targeting this stage in BM development through administration of apomorphine. The ability to prevent metastatic progression to the brain can transform an unvaryingly lethal systemic disease into one that is eminently more treatable.

Patient sample collection and cell culture

Human lung-derived BM were obtained with written consent from patients, as approved by the Hamilton Health Sciences/McMaster Health Sciences Research Ethics Board (REB # 07366), in compliance with Canada's Tri-Council Policy Statement on the Ethical Conduct for Research Involving Humans and the International Ethical Guidelines for Biomedical Research Involving Human Subjects.

BMs were processed and maintained in tumor sphere media (TSM) as previously described (14, 15). BMICs were grown as tumorspheres that were maintained at 37°C with a humidified atmosphere of 5% CO2.

In vivo modeling of metastasis

All experimental procedures involving animals were reviewed and approved by McMaster University Animal Research Ethics Board. NOD-SCID mice were used for all experiments. Mice were anesthetized using gas anesthesia (isoflurane: 5% induction, 2.5% maintenance) before minimally invasive surgery. Injections were performed as previously described for intracranial (ICr), intrathoracic (IT), and intracardiac (ICr) routes (Supplementary Table S1A; ref. 14).

Mice were monitored weekly, and upon reaching endpoint brains and lungs were harvested and underwent two separate analyses:

  1. Hemotoxylin and eosin staining. For ICr injections, 100,000 cells of BT478 (n = 2) and BT530 (n = 2) were utilized, for ICa injections 250,000 cells of BT478 (n = 6) and BT530 (n = 2), and for IT injections 500,000 cells of BT478 (n = 7) and BT530 (n = 2). Whole brains (and lungs from IT injections) were sectioned and paraffin-embedded for hematoxylin and eosin. Images were scanned using an Aperio Slide Scanner and analyzed by ImageScope v11.1.2.760 software (Aperio).

  2. In vitro culture and expansion. For ICr injections, 50,000 cells of BT478 (n = 3) and BT530 (n = 4) were utilized, for ICa injections 250,000 cells of BT478 (n = 9) and BT530 (n = 6), and for IT injections 500,000 cells of BT478 (n = 17) and BT530 (n = 9). BMICs were reisolated from ICr brain tumors (BT), IT lung tumors (LT), and premetastatic brain tumors (BMIT), and ICa brain tumors (BMIC) as follows: Whole brains and lungs (IT injections) were dissociated into single cell suspensions (15) and cultured in DMEM with decreasing concentrations of FBS—the first 2 days in 20% FBS, 10% FBS for 2 to 3 days, 5% FBS, and finally in TSM with puromycin for a minimum of 1 week prior to any analyses to select out any residual contamination of mouse cells as well as to enrich for the BMICs. Duplicate samples per BT, LT, BMIT, and BMIC were collected per BMIC line, RNA isolated, and submitted for microarray analyses (BT478) or RNA sequencing analyses (BT478 and BT530).

For drug treatments, mice were injected through IT (control, n = 5; apo tx, n = 10) and IC route (control, n = 5; apo tx, n = 10), and cells were allowed to engraft for 2 weeks. R-(−)-Apomorphine hydrochloride hemihydrate (Sigma) was resuspended in sterile saline at 0.5 mg/mL and administered by subcutaneous (s.c.) injections to give a final dose of 5 mg/kg, 3 times weekly for 1 month. Control mice received only saline. Control mice were culled as they succumbed to endpoint, and 2 corresponding apomorphine treatment mice were culled for each control mouse to complete a matched set.

IC50 curve generation

BMICs were dissociated into a single cell suspension, and 2,000 cells/well were plated into a 96-well plate at a volume of 200 mL/well in increasing concentrations (5–25 μmol/L) of apomorphine, GW-8510, lomustine (Sigma), acacetin (Sigma), thioridazine (Sigma), trifluoroperazine (Sigma), and prochlorperazine (Sigma). DMSO was used as a control. Cells were incubated for 4 days. Presto Blue (20 μL; Invitrogen) was added to each well approximately 2 hours prior to the readout time point. Fluorescence was measured using a FLUOstar Omega Fluorescence 556 Microplate reader (BMG LABTECH) at excitation and emission wavelengths of 535 nm and 600 nm, respectively. Readings were analyzed using Omega analysis software. Dose–response curves were fitted to the data.

Reverse transcription and quantitative PCR of mRNA

Total RNA was isolated using Norgen RNA extraction kit (Biotek) and reverse transcribed using qScript cDNA Super Mix (Quanta Biosciences) and a C1000 Thermo Cycler (Bio-Rad). qRT-PCR was performed using the Cfx96 (Bio-Rad) with SsoAdvanced SYBR Green (Bio-Rad) using gene specific primers (Supplementary Table S2) and GAPDH as the internal control.

Flow-cytometric characterization

Adherent BMICs were detached through application of TryplE (Invitrogen) and single cells were resuspended in PBS+2 mmol/L EDTA. Cell suspensions were stained with human anti-TRA-1-85 (CD147, Miltenyi) and incubated for 30 minutes on ice. Samples were run on a MoFlo XDP Cell Sorter (Beckman Coulter). Dead cells were excluded using the viability dye 7AAD (1:10; Beckman Coulter). Compensation was performed using mouse IgG CompBeads (BD). Surface marker expression was defined as positive or negative based on the analysis regions established using the isotype control.

Microarray data analyses

BT478 samples were prepared, processed, and run as per Illumina protocol as previously described (16). Illumina summary probe profiles along with associated control probe profiles were read using a Bioconductor package limma v3.30.13 (17). Data were then background corrected using negative control probes and subsequently normalized applying quantile normalization using all the available control probes. After normalization, expression of the genes was averaged across the technical replicates obtained from the same biological sample.

To provide qualitative assessment of the dissimilarity of the BMIT against BT, LT, and BMIC, scatterplots were plotted depicting expression of the genes as obtained from individual samples. The Pearson coefficient of correlation between the individual samples was calculated and plotted to generate a heat map of the obtained correlations.

RNA sequencing

Illumina sequencing was performed by the Farncombe Metagenomics Facility (McMaster University). RNA integrity was first verified using the Agilent BioAnalyzer, followed by mRNA enrichment and library prep using the NEBNext Ultra Directional RNA Library Prep Kit along with the NEBNext Poly(A) mRNA Magnetic Isolation Module. Libraries were subject to further BioAnalyzer QC and quantified by qPCR. Sequencing was performed using the HiSeq Rapid v2 chemistry with paired end 2 × 50 bp read length configurations.

Raw RNA sequencing data were preprocessed and normalized as follows: RNA-seq data were aligned against hg38 reference genome, using bowtie2. Reads counts per gene were obtained using R packages GenomicRange and GenomicFeatures and using UCSC hg38 KnowGene database as a reference for genomic locations (TxDb.Hsapiens.UCSC.hg38). Counts were first normalized to counts per million, and then additional quantile normalization was applied. Expressions were averaged across pairs of technical replicates. Counts were then log2 transformed, and genes whose expression was <0 across all the 18 samples were removed. Principal component analysis (PCA) was then conducted and all the samples were depicted in the space defined by the two most principal components. Additionally, a heat map depicting sample differences, as quantified by Euclidean distance of the gene expression, was generated along the dendrogram depicting hierarchical clustering of the samples. Sample 16 was then excluded from further analysis as an outlier.

Differential expression analysis was performed to identify genes whose expression was significantly different when comparing: (i) BMIT against BT, LT, and BMIC from BT478 and (ii) BMIT against BT, LT, and BMIC of BT530. Using Bioconductor package limma v3.30.13 (17), log2-fold change of the gene expression was calculated for both comparisons along the associated P value and false discovery rate.

Enrichment analysis

Two types of enrichment analysis were conducted, gene set enrichment analysis (GSEA) as described by Subramanian and colleagues (18), along with overrepresentation analysis using hypergeometric test to assess significance of overlap between the selected group of genes and given pathway or biological process. In both cases, enrichment against the 5 major ontologies was assessed, including Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (19), Gene Ontology (GO)–biological processes, GO–cellular components, GO–molecular functions (20), and disease ontology (DO; ref. 21). All the enrichment analyses were performed using functions implemented within the Bioconductor package ClusterProfiler v3.2.14 (22).

CMAP analysis

CMAP analysis was used to predict effects of the drugs on the expression of the deregulated genes (23). In this analysis, drugs (comprising 1,289 chemical substances) were assessed with respect to their ability to invert expression changes of the deregulated genes obtained from above described differential gene expression analysis. CMAP analysis was conducted using Bioconductor package PharmacoGx (24). Drugs were first filtered according to resulting connectivity score (connectivity score < 0) and associated significance (P < 0.01). Finally, drugs were selected for preliminary in vitro screening based on the criteria of novelty in metastasis treatment, ability to cross the blood–brain barrier, and potential to target neural developmental systems or associated disorders.

To further explore effects of apomorphine on gene expression, we constructed a protein–protein interaction (PPI) network using apomorphine gene targets obtained from DrugBank v5.0.11 (25) and The Comparative Toxicogenomics Database vJan-2018 (26). Genes transcriptionally modified by apomorphine were identified using CMAP ver. 1 (23). Using the three gene lists, we then identified PPIs connecting individual genes in the list using Integrated Interactions Database IID v2017-04 (27). The resulting PPI network was visualized using NAViGaTOR v3 (28). As per legend, node color represents GO–Molecular Function; edge color corresponds to tissue specificity, specifically highlighting lung and brain tissue, as obtained from IID. The most important BMIT gene targets of apomorphine were identified by applying PharmacoGx framework for sensitivity modeling (for more details see PharmacoGx user's guide). Genes were filtered according to the drug's estimated effect on their expression (upregulation of the downregulated genes and downregulation of the upregulated ones) and associated significance (P < 0.01).

Kaplan–Meier analysis

Prognostic potential of the genes targeted by the selected drugs was assessed through SurvExpress v2.0—web resource for validation of cancer gene expression biomarkers (http://bioinformatica.mty.itesm.mx:8080/Biomatec/SurvivaX.jsp; ref. 29) and lung module of Kaplan–Meier plotter (KMplotter)—tool for meta-analysis-based biomarker assessment (http://kmplot.com; ref. 30). Prognostic significance of the 3 target genes (KIF16B, SEPW1, and TESK2) was first tested in SurvExpress using The Cancer Genome Atlas (TCGA) lung adenorcarcinoma gene expression data set (June 2016) and then validated in KMplotter using all available lung adenocarcinoma data sets. In both cases, survival analysis was conducted under default parametrization.

Statistical analysis

Replicates from at least 3 samples were used for IC50 and RT-PCR experiments. Respective data represent mean ± SD with n values listed in figure legends. Student t test and two-way ANOVA analyses were conducted using GraphPad Prism 5. P < 0.05 was considered statistically significant.

Data availability

The authors declare that all the Supplementary Data of the findings of this study are available within the article, its Supplementary information files, and from the corresponding author upon reasonable request. RNA sequencing files are available as GEO data set GSE110495 at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE110495 upon request.

Capturing the pre- and macrostages of metastatic growth in BM development

We utilized early-passage BM cell lines derived from primary patient samples of lung-to-BM in our work, as these samples are enriched for BMICs that have already successfully completed the metastatic process. Previous work in our lab successfully established preclinical models of lung-to-brain BM (14, 31). Briefly, we injected mice through 3 different injection routes: (i) intracranial (ICr), (ii) intrathoracic injections (IT), and (iii) intracardiac injections (ICa), where we were able to replicate the premetastatic and macrometastatic stages from IT and ICa injections respectively (14). Here, we have further isolated and characterized BMICs at each metastatic stage. BMIC lines transduced with GFP were injected into our BM models and were shown to reform tumors at each stage of the metastatic cascade, from primary lung (LT) and secondary brain (BT) tumor formation to the premetastatic (BMIT) and macrometastasis (BMIC) stages of tumor growth (Fig. 1). Approximate timeframes for tumor development (endpoint) varied between models and cell line injection (Supplementary Table S1A); however, there was approximately 10 to 14 days difference between ICa and IT endpoints. BMICs were isolated from BT, BMIT, and BMIC tumors and minimally cultured, and retained the ability to reform secondary spheres, suggesting a preservation of their stem-like and tumor initiation properties (Fig. 1).

Figure 1.

Isolation and characterization of in vivo BMICS, BT, LT, BMIT, and BMICs. Top, BT478 and BT530 BMICs were tagged with a GFP-expressing vector containing a puromycin-resistant cassette. GFP+ BMICs were injected via ICr, ICa, and IT routes and characterized via hematoxylin and eosin staining. BMICs are able to recapitulate metastatic stages of primary lung (LT) and secondary orthotopic brain (BT) tumors, micrometastases (BMIT) and macrometastases (BMIC). Bottom, whole organs (brain or lung) were isolated from each metastatic stage and cultured under TSM conditions with puromycin to select for only GFP+ BMICs, where recovered BMICs were able to reform spheres. Scale bar, 400 μm.

Figure 1.

Isolation and characterization of in vivo BMICS, BT, LT, BMIT, and BMICs. Top, BT478 and BT530 BMICs were tagged with a GFP-expressing vector containing a puromycin-resistant cassette. GFP+ BMICs were injected via ICr, ICa, and IT routes and characterized via hematoxylin and eosin staining. BMICs are able to recapitulate metastatic stages of primary lung (LT) and secondary orthotopic brain (BT) tumors, micrometastases (BMIT) and macrometastases (BMIC). Bottom, whole organs (brain or lung) were isolated from each metastatic stage and cultured under TSM conditions with puromycin to select for only GFP+ BMICs, where recovered BMICs were able to reform spheres. Scale bar, 400 μm.

Close modal

To characterize the genetic profiles of each stage of metastatic progression, we performed preliminary microarray analysis of BT478 BMICs from BT, LT, BMIT, and BMIC samples. Intriguingly, we found that genes from BMIT cells clustered separately from BT, LT, and BMIC samples (Supplementary Fig. S1A and S1B). To corroborate this unique premetastatic BMIT genetic profile, we analyzed RNA sequencing data obtained across two separate BMIC lines. Hierarchical clustering along PCA showed that BMIT from both BMIC lines cluster together, irrespective of the cell line origin, whereas established metastatic tumors (BT, LT, BMIC) group into cell line–specific clusters (Fig. 2A and B; Supplementary Fig. S1C). We then performed differential expression analysis comparing expression profiles of BMIT with non-BMIT samples from both cell lines separately. We identified ∼7,000 differentially expressed genes in the premetastatic BMIT stage (Supplementary Data Set S1). These results indicate temporal evolution of BMICs through metastasis, during which a distinct genetic profile emerges prior to the initiation of the secondary brain metastasis, while established tumors retain a genetically similar profile despite tissue of origin.

Figure 2.

Characterization of the individual stages of brain metastasis progression. A, Heat maps depicting Pearson correlation coefficient of gene expression across the samples as measured initially by RNA-seq, along with associated hierarchical clustering of the samples using Euclidean distance between samples expression profiles. B, PCA plot depicting samples in the plane defined by two main components (percentage indicates variance explained; “original” denotes BMIC samples collected prior to injection).

Figure 2.

Characterization of the individual stages of brain metastasis progression. A, Heat maps depicting Pearson correlation coefficient of gene expression across the samples as measured initially by RNA-seq, along with associated hierarchical clustering of the samples using Euclidean distance between samples expression profiles. B, PCA plot depicting samples in the plane defined by two main components (percentage indicates variance explained; “original” denotes BMIC samples collected prior to injection).

Close modal

Premetastatic BMICs retain a unique genetic profile

Using GSEA, we assessed association of BMIT deregulated genes with biological processes (GO), cellular components (GO), molecular functions (GO), biological pathways (KEGG) or diseases (DO). We found increased expression of genes regulating cytoskeletal structures and epithelial tumor invasion, as well as decreased expression in processes of cell division and apoptosis (Fig. 3A and B; Supplementary Data Set S2). These data suggest that premetastatic BMIT are not dormant, but have concurrently increased activation of invasive mechanisms while repressing programmed cell death and growth mechanisms. We also found enrichment within several neurodegenerative pathways (Supplementary Fig. S2; Supplementary Data Set S3) and neural neoplasm components (Supplementary Fig. S3, Supplementary Data Set S4). We also performed enrichment analysis (overrepresentation analysis) of the gene clusters obtained by hierarchical clustering of BT, LT, BMIT, and BMIT genes (Fig. 3C). We identified clusters of BMIT deregulated genes to be significantly (P < 0.01) enriched in pathways of cancer and neuroactive ligand–receptor interaction. Interestingly, enrichment analysis of the instances of the DO revealed enrichment of the autonomic nervous system neoplasm (Supplementary Data Set S5).

Figure 3.

Cellular processes and biological pathways associated with BMIT. A, Visualization of the GSEA across GO–cellular components ontology and KEGG pathways database, using BMIT deregulated genes ordered according to their expression fold change [y-axis, statistical significance; point size, size of the gene set (cellular component/pathway); color, normalized enrichment score (NES)]. B, Heat maps depicting Pearson correlation coefficient of gene expression in select cellular processes across the samples as measured initially by RNA sequencing. C, Heat map depicting expression of the BMIT-deregulated genes across all the samples along the dendrogram obtained by hierarchical clustering of these genes. Enrichment (overrepresentation) analysis of BMIT genes across individual branches of the dendrogram revealed enrichment of several KEGG pathways, as well as DO instances, GO biological processes, cellular compartments, and molecular functions (“original” denotes BMIC samples collected prior to injection).

Figure 3.

Cellular processes and biological pathways associated with BMIT. A, Visualization of the GSEA across GO–cellular components ontology and KEGG pathways database, using BMIT deregulated genes ordered according to their expression fold change [y-axis, statistical significance; point size, size of the gene set (cellular component/pathway); color, normalized enrichment score (NES)]. B, Heat maps depicting Pearson correlation coefficient of gene expression in select cellular processes across the samples as measured initially by RNA sequencing. C, Heat map depicting expression of the BMIT-deregulated genes across all the samples along the dendrogram obtained by hierarchical clustering of these genes. Enrichment (overrepresentation) analysis of BMIT genes across individual branches of the dendrogram revealed enrichment of several KEGG pathways, as well as DO instances, GO biological processes, cellular compartments, and molecular functions (“original” denotes BMIC samples collected prior to injection).

Close modal

Therapeutic targeting of premetastatic BMIT

CMAP was performed on the dysregulated BMIT gene set to identify potential targeting therapeutics (Supplementary Table S3; Supplementary Data Set S6). Drugs were selected for preliminary in vitro screening based on the criteria of novelty in metastasis treatment, ability to cross the blood–brain barrier, and potential targeting of neural developmental systems or associated disorders, from which the DRD2 agonist apomorphine proved to have a moderately low IC50 for both BT478 and BT530 BMIC lines (Fig. 4). We repeated the drug screening with other dopamine-specific psychological therapeutics, which failed to affect BMICs to the same extent as apomorphine (Fig. 4).

Figure 4.

In vitro IC50 screening of potential brain metastasis targeting drugs. IC50 curves of selected BMIT-targeted drugs. (n = 3).

Figure 4.

In vitro IC50 screening of potential brain metastasis targeting drugs. IC50 curves of selected BMIT-targeted drugs. (n = 3).

Close modal

To assess the efficacy of apomorphine inhibiting BMITin vivo, we performed ICa injections with BMIC line BT478, following a modified protocol utilized for in vivo Alzheimer models treated with apomorphine (32). BMICs were allowed to engraft for 2 weeks prior to starting a month-long administration of apomorphine, 3 times weekly along with saline for control mice (Fig. 5A). Despite apomorphine being a known emetic, the treated mice displayed no significant weight loss, whereas there was a slight decrease in control mice weights (Supplementary Table S1B). Mice were culled at endpoint (approximately 2.5 months posttumor injection for ICa, and 2 months posttumor injection for IT), and their brains minimally cultured to remove the bulk of mouse cellular debris. We then performed FACS for human-Tra-1-85 to isolate human BMICs. apomorphine greatly attenuated BM development through the ICa BM model, as defined by a complete absence of BMICs in apomorphine-treated brains (Fig. 5B; Supplementary Fig. S4), suggesting that apomorphine does target BMIT cells to prevent BM initiation and development, both in silico and in vivo. The efficacy of apomorphine in inhibiting BM development in the IT model was indeterminable, as the relatively low number of BMICs that were reisolated from both the control and apomorphine-treated mice made it difficult to determine a difference (Supplementary Fig. S5A).

Figure 5.

Preclinical testing of apomorphine to prevent brain metastasis. A, Schematic representation of dosing regimen for apomorphine. B, Scatter plot graph depicting percentage of human-Tra-1-85–positive GFP-tagged BMIC cells reisolated from apomorphine (Apo) treatment and control (CNTL) ICa BM model (control, n = 3; treatment, n = 6; ****, P < 0.0001).

Figure 5.

Preclinical testing of apomorphine to prevent brain metastasis. A, Schematic representation of dosing regimen for apomorphine. B, Scatter plot graph depicting percentage of human-Tra-1-85–positive GFP-tagged BMIC cells reisolated from apomorphine (Apo) treatment and control (CNTL) ICa BM model (control, n = 3; treatment, n = 6; ****, P < 0.0001).

Close modal

Premetastatic BMIT genes are predictive of lung cancer patient survival

We attempted to elucidate the biological context of apomorphine to determine possible mechanisms of actions. We first generated an interactome to identify overall genes targeted by apomorphine (Fig. 6A). Application of a targeted PharmacoGx framed CMAP on apomorphine focusing on the premetastatic BMIT genes identified 3 genes downregulated as direct targets, KIF16B, SEPW1, and TESK2 (Fig. 6B). In vitro analyses determined transcript levels of these 3 genes to be moderately increased in BMICs treated with apomorphine (Fig. 6C). These 3 genes were then interrogated for prognostic value using transcriptomic data from a lung adenocarcinoma patient cohort. The genes taken individually as well as a refined collective signature comprised of TESK2, SEPW1, and KIF16B were found to have significant impact on patient survival, where low expression of these genes correlated with poor patient survival (Fig. 6C; Supplementary Fig. S5B).

Figure 6.

Novel gene targets of apomorphine. A, PPI network identifying common gene targets of apomorphine. B, BMIT genes directly targeted by apomorphine, as determined by CMAP analysis; negative direction values depict low gene expression, which is correlated with poor prognosis. Relative transcript levels of KIF16B, SEPW1, and TESK2 in BMICs treated with apomorphine (n = 3; ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001). C, Kaplan–Meier curves depicting expression of apomorphine-targeted genes by risk group as obtained from SurvExpress using TCGA data from patients with lung adenocarcinoma.

Figure 6.

Novel gene targets of apomorphine. A, PPI network identifying common gene targets of apomorphine. B, BMIT genes directly targeted by apomorphine, as determined by CMAP analysis; negative direction values depict low gene expression, which is correlated with poor prognosis. Relative transcript levels of KIF16B, SEPW1, and TESK2 in BMICs treated with apomorphine (n = 3; ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001). C, Kaplan–Meier curves depicting expression of apomorphine-targeted genes by risk group as obtained from SurvExpress using TCGA data from patients with lung adenocarcinoma.

Close modal

Our limited mechanistic understanding of metastatic disease greatly hinders therapeutic discovery and improvement of the dismal patient outcome of BM (33). Despite advancements in preventative and treatment modalities for primary tumors that have resulted in increased patient survival, the inability of these treatments to target residual CSC and BMIC populations leaves patients with cancer vulnerable and prone to relapse and metastases (34).

Significant study of the genome evolution of cancer has identified precancerous events in several primary cancers (7, 8, 35); unfortunately, the molecular mechanisms that drive premetastatic cells in the brain remain poorly defined. A significant disadvantage with currently available in vivo models is the inability to capture the premetastatic stage of brain tissue colonization, instead focusing on the easier to collect macrometastatic stage. Recent studies with C. elegans led by Matus and colleagues (12) determined that cellular invasion and proliferation are mutually incompatible stages, where both stages are representative of premetastasis and macrometastasis progression, respectively. This work substantiates the inefficient targeting of invasive cells by current chemotherapies that tend to target rapidly dividing cells, perhaps at the expense of invasive cells (36).

Previous work in our lab successfully established clinically relevant models of lung-derived BM representing the different stages of metastasis, where we captured both the premetastatic and macrometastatic stages of tumor growth via our IT and IC routes, respectively (14). From our intrathoracic BM model, we found that mice characteristically die of lung tumor burden just as BMICs cross the blood–brain barrier and colonize the brain, giving us a time point to isolate these premetastatic BMICs. Through isolation and comparison of BMICs at various stages of metastatic progression in our established BM models, we identified a genetic pattern unique only to BMICs undergoing premetastasis, termed BMIT, whereas established macrometastatic tumors (BT, LT, and BMIC) were genetically similar. These BMIT-BMICs possess ∼7,000 dysregulated genes, active in mechanisms that promote invasion and repress apoptosis and division, corroborating results by Matus and colleagues in our more relevant patient-related modeling systems (12). Where the use of NOD-SCID mice encourages increased rates of engraftment of patient BMIC lines, it is possible that the lack of a full immune system does not provide information on the full scope of metastatic progression. Current studies concerning the interaction of the immune system and metastatic cells suggest an intricate relationship, where immune cells can mediate metastatic cell entry into the CNS as well as modulate BM growth (37). The addition of an active immune system may likely reduce the rate of BMIC engraftment in our BM models, possibly requiring inoculation of higher cell numbers or longer incubation times to tumor development.

The role of neurotransmitters in cancer has drawn varying interest over the years, where they have been found to exert a strong influence over external and internal cellular factors in cancer progression (38). Breast cancer BMICs have been found to exhibit GABAergic properties, mimicking neuronal phenotypes that appear to aid their colonization of the brain (39). Dopamine receptors (DR) and dopamine have been revealed to exhibit various pleiotropic properties through dependent and independent pathways, and their modulation has enhanced the efficiency of anticancer drugs in preclinical cancer models (40, 41). In particular, DRD2 agonists have recently been shown to suppress proliferation, angiogenesis, and invasion in several cancers and tumors (42–44). Such studies paired with epidemiologic data implicate a relationship between lower rates of cancer development in patients with Parkinson, intimating a possible link between DR agonists and cancer (45, 46).

Through enrichment analyses, we determined that BMIT dysregulated gene sets enrich pathways that regulate autonomic nervous system neoplasms and neural system dysregulation, implying a possible relation between neurodevelopmental pathways and promotion of cancer invasion. CMAP interrogation of the dysregulated BMIT genes identified a list of targeting therapeutics, of which several of the top hits are currently applied, or are being investigated, as antineoplastic agents against various cancers (47–49). We selected drugs for preliminary in vitro screening based on the ability to pass the BBB, treatment of neurologic disorders, and overall novelty as a cancer therapeutic, from which apomorphine was selected for further validation. Apomorphine is a nonselective dopamine agonist of the morphine derivative, primarily activating dopamine-like receptor 2 (DRD2). Among its multiple uses, apomorphine administration reduced amyloid β degradation in patients with Alzheimer (32) and has recently shown efficacy in the treatment of Parkinson disease (50) as well as a potential targeting of tumor cell invasion (51). Further screening against other dopamine-specific psychological therapeutics validated the specific efficacy of apomorphine in targeting premetastatic BMICs.

To further validate the ability of apomorphine to target BMIT, we applied the drug in vivo in our BM models. Initial trials administering apomorphine against our IT model drew inconclusive results, where the relatively low number of BMICs we were able to capture at the premetastatic stage made it difficult to confidently determine the efficacy of apomorphine (Supplementary Fig. S6A). Thus, we utilized our ICa model to properly interrogate the efficacy of apomorphine against BM development, collecting samples at early time points that follow the micrometastatic time course of our IT model as well as at survival endpoint to confirm macrometastatic growth. Apomorphine proved to be successful at inhibiting micrometastatic growth as well as subsequent macrometastases. We utilized a treatment protocol modified from in vivo Alzheimer models being treated with apomorphine, as these models proved apomorphine to be effective and tolerable at the administered dosages. However, future studies will look to tailor the apomorphine dosage to determine the lowest concentration for BM inhibition.

PharamcoGx directed CMAP analysis determined 3 downregulated BMIT genes specifically targeted by apomorphine—KIF16B, SEPW1, and TESK2—where in silico application of the drug would activate their expression. SEPW1 belongs to a family of selenoproteomes, which have been increasingly implicated in aspects of neurobiology and neurodegenerative disorders (52). TESK2 is a serine/threonine protein kinase (53). KIF16B is a kinesin-like motor protein that may be involved in intracellular trafficking (54), where defects in this family of proteins has been associated with neurodegenerative, developmental, and cancer diseases (55). In vitro analysis of apomorphine-treated BMICs determined transcript levels of these 3 genes to be moderately increased as compared with the control. When these genes were applied both individually and as a collective signature in a cohort of patients with lung adenocarcinoma, their low expression was correlated with poorer patient survival. Further interrogation of data that follow patients with lung cancer progression into BM development will be required to validate the predictive value of TESK2, SEPW1, and KIF16B. It is anticipated that, with the discovery of our novel premetastatic gene set, we could predict or identify the potential for metastasis in either primary lung cancer or circulating tumor cells prior, thus any treatment to be administered would be on a preventative basis and hopefully circumvent the need for the current dismal treatment options. We are well aware that any therapeutic administered could alter the nature of the tumor and promote metastasis through a resistant population; however, we are optimistic that our preventative treatment would extend patient survival long enough to determine an alternative treatment if necessary.

We present an in-depth genetic characterization of the previously uncaptured stage of premetastasis in BM progression. We further identified apomorphine to be a novel BMIT targeting therapeutic to prevent BM development. Continuing studies will further characterize the role and related mechanisms of DR agonists in BM development. The ability to inhibit BMICs from initiating metastasis would target BM at the ideal stage, preventing the need for more toxic and possibly detrimental treatments. Our identification of this premetastatic stage in the development of BM can be mined to provide further critical therapeutic targets in all cancers that metastasize to the brain, offering a paradigm shift for the current state of BM treatment.

No potential conflicts of interest were disclosed.

Conception and design: M. Singh, C. Venugopal, T. Tokar, S.K. Singh

Development of methodology: M. Singh, C. Venugopal, T. Tokar, M.K. Subapanditha, M. Qazi, D. Bakhshinyan, S.K. Singh

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.K. Subapanditha, M. Qazi, P. Vora, N.K. Murty, S.K. Singh

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Singh, C. Venugopal, T. Tokar, I. Jurisica, S.K. Singh

Writing, review, and/or revision of the manuscript: M. Singh, C. Venugopal, T. Tokar, I. Jurisica, S.K. Singh

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N. McFarlane

Study supervision: C. Venugopal, I. Jurisica, S.K. Singh

M. Singh was supported by the Brain Canada PhD Studentship. This work was supported by funds from the Department of Surgery at McMaster University, Canadian Cancer Society Innovation to Impact Grant (i2I16-1) and The Boris Family Fund for Brain Metastasis Research awarded to S.K. Singh, and Ontario Research Fund (GL2-01-030), Canada Research Chair Program (CRC #225404), and Canada Foundation for Innovation (CFI #29272, #225404, #30865) awarded to I. Jurisica.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Patchell
RA
. 
The management of brain metastases
.
Cancer Treat Rev
2003
;
29
:
533
40
.
2.
Sjobakk
TE
,
Vettukattil
R
,
Gulati
M
,
Gulati
S
,
Lundgren
S
,
Gribbestad
IS
, et al
Metabolic profiles of brain metastases
.
Int J Mol Sci
2013
;
14
:
2104
18
.
3.
Chambers
AF
,
Groom
AC
,
MacDonald
IC
. 
Dissemination and growth of cancer cells in metastatic sites
.
Nat Rev Cancer
2002
;
2
:
563
72
.
4.
Vanharanta
S
,
Massague
J
. 
Origins of metastatic traits
.
Cancer Cell
2013
;
24
:
410
21
.
5.
Brastianos
PK
,
Carter
SL
,
Santagata
S
,
Cahill
DP
,
Taylor-Weiner
A
,
Jones
RT
, et al
Genomic characterization of brain metastases reveals branched evolution and potential therapeutic targets
.
Cancer Discov
2015
;
5
:
1164
77
.
6.
Anderson
K
,
Lutz
C
,
van Delft
FW
,
Bateman
CM
,
Guo
Y
,
Colman
SM
, et al
Genetic variegation of clonal architecture and propagating cells in leukaemia
.
Nature
2011
;
469
:
356
61
.
7.
Hong
D
,
Gupta
R
,
Ancliff
P
,
Atzberger
A
,
Brown
J
,
Soneji
S
, et al
Initiating and cancer-propagating cells in TEL-AML1-associated childhood leukemia
.
Science
2008
;
319
:
336
9
.
8.
Shlush
LI
,
Zandi
S
,
Mitchell
A
,
Chen
WC
,
Brandwein
JM
,
Gupta
V
, et al
Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia
.
Nature
2014
;
506
:
328
33
.
9.
Wistuba
II
,
Mao
L
,
Gazdar
AF
. 
Smoking molecular damage in bronchial epithelium
.
Oncogene
2002
;
21
:
7298
306
.
10.
Celia-Terrassa
T
,
Kang
Y
. 
Distinctive properties of metastasis-initiating cells
.
Genes Dev
2016
;
30
:
892
908
.
11.
Massague
J
,
Obenauf
AC
. 
Metastatic colonization by circulating tumour cells
.
Nature
2016
;
529
:
298
306
.
12.
Matus
DQ
,
Lohmer
LL
,
Kelley
LC
,
Schindler
AJ
,
Kohrman
AQ
,
Barkoulas
M
, et al
Invasive cell fate requires G1 cell-cycle arrest and histone deacetylase-mediated changes in gene expression
.
Dev Cell
2015
;
35
:
162
74
.
13.
Klus
P
,
Cirillo
D
,
Botta Orfila
T
,
Gaetano Tartaglia
G
. 
Neurodegeneration and cancer: where the disorder prevails
.
Sci Rep
2015
;
5
:
15390
.
14.
Singh
M
,
Venugopal
C
,
Tokar
T
,
Brown
KR
,
McFarlane
N
,
Bakhshinyan
D
, et al
RNAi screen identifies essential regulators of human brain metastasis-initiating cells
.
Acta Neuropathol
2017 134:923–40
.
15.
Venugopal
C
,
McFarlane
NM
,
Nolte
S
,
Manoranjan
B
,
Singh
SK
. 
Processing of primary brain tumor tissue for stem cell assays and flow sorting
.
J Vis Exp
2012
.
16.
Venugopal
C
,
Hallett
R
,
Vora
P
,
Manoranjan
B
,
Mahendram
S
,
Qazi
MA
, et al
Pyrvinium targets CD133 in human glioblastoma brain tumor-initiating cells
.
Clin Cancer Res
2015
;
21
:
5324
37
.
17.
Ritchie
ME
,
Phipson
B
,
Wu
D
,
Hu
Y
,
Law
CW
,
Shi
W
, et al
limma powers differential expression analyses for RNA-sequencing and microarray studies
.
Nucleic Acids Res
2015
;
43
:
e47
.
18.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
19.
Kanehisa
M
,
Goto
S
. 
KEGG: Kyoto encyclopedia of genes and genomes
.
Nucleic Acids Res
2000
;
28
:
27
30
.
20.
Ashburner
M
,
Ball
CA
,
Blake
JA
,
Botstein
D
,
Butler
H
,
Cherry
JM
, et al
Gene ontology: tool for the unification of biology. The gene ontology consortium
.
Nat Genet
2000
;
25
:
25
9
.
21.
Schriml
LM
,
Arze
C
,
Nadendla
S
,
Chang
YW
,
Mazaitis
M
,
Felix
V
, et al
Disease ontology: a backbone for disease semantic integration
.
Nucleic Acids Res
2012
;
40
(
Database issue
):
D940
6
.
22.
Yu
G
,
Wang
LG
,
Han
Y
,
He
QY
. 
clusterProfiler: an R package for comparing biological themes among gene clusters
.
OMICS
2012
;
16
:
284
7
.
23.
Lamb
J
. 
The connectivity map: a new tool for biomedical research
.
Nat Rev Cancer
2007
;
7
:
54
60
.
24.
Smirnov
P
,
Safikhani
Z
,
El-Hachem
N
,
Wang
D
,
She
A
,
Olsen
C
, et al
PharmacoGx: an R package for analysis of large pharmacogenomic datasets
.
Bioinformatics
2016
;
32
:
1244
6
.
25.
Wishart
DS
,
Feunang
YD
,
Guo
AC
,
Lo
EJ
,
Marcu
A
,
Grant
JR
, et al
DrugBank 5.0: a major update to the DrugBank database for 2018
.
Nucleic Acids Res
2018
;
46
(
D1
):
D1074
D82
.
26.
Davis
AP
,
Grondin
CJ
,
Johnson
RJ
,
Sciaky
D
,
King
BL
,
McMorran
R
, et al
The comparative toxicogenomics database: update 2017
.
Nucleic Acids Res
2017
;
45
(
D1
):
D972
D8
.
27.
Kotlyar
M
,
Pastrello
C
,
Sheahan
N
,
Jurisica
I
. 
Integrated interactions database: tissue-specific view of the human and model organism interactomes
.
Nucleic Acids Res
2016
;
44
(
D1
):
D536
41
.
28.
Brown
KR
,
Otasek
D
,
Ali
M
,
McGuffin
MJ
,
Xie
W
,
Devani
B
, et al
NAViGaTOR: network analysis, visualization and graphing Toronto
.
Bioinformatics
2009
;
25
:
3327
9
.
29.
Aguirre-Gamboa
R
,
Gomez-Rueda
H
,
Martinez-Ledesma
E
,
Martinez-Torteya
A
,
Chacolla-Huaringa
R
,
Rodriguez-Barrientos
A
, et al
SurvExpress: an online biomarker validation tool and database for cancer gene expression data using survival analysis
.
PLoS One
2013
;
8
:
e74250
.
30.
Gyorffy
B
,
Surowiak
P
,
Budczies
J
,
Lanczky
A
. 
Online survival analysis software to assess the prognostic value of biomarkers using transcriptomic data in non-small-cell lung cancer
.
PLoS One
2013
;
8
:
e82241
.
31.
Nolte
SM
,
Venugopal
C
,
McFarlane
N
,
Morozova
O
,
Hallett
RM
,
O'Farrell
E
, et al
A cancer stem cell model for studying brain metastases from primary lung cancer
.
J Natl Cancer Inst
2013
;
105
:
551
62
.
32.
Himeno
E
,
Ohyagi
Y
,
Ma
L
,
Nakamura
N
,
Miyoshi
K
,
Sakae
N
, et al
Apomorphine treatment in Alzheimer mice promoting amyloid-beta degradation
.
Ann Neurol
2011
;
69
:
248
56
.
33.
Steeg
PS
. 
Targeting metastasis
.
Nat Rev Cancer
2016
;
16
:
201
18
.
34.
Langley
RR
,
Fidler
IJ
. 
The biology of brain metastasis
.
Clin Chem
2013
;
59
:
180
9
.
35.
Sgroi
DC
. 
Preinvasive breast cancer
.
Annu Rev Pathol
2010
;
5
:
193
221
.
36.
Hurst
RE
,
Bastian
A
,
Bailey-Downs
L
,
Ihnat
MA
. 
Targeting dormant micrometastases: rationale, evidence to date and clinical implications
.
Ther Adv Med Oncol
2016
;
8
:
126
37
.
37.
Hamilton
A
,
Sibson
NR
. 
Role of the systemic immune system in brain metastasis
.
Mol Cell Neurosci
2013
;
53
:
42
51
.
38.
Jobling
P
,
Pundavela
J
,
Oliveira
SM
,
Roselli
S
,
Walker
MM
,
Hondermarck
H
. 
Nerve-cancer cell cross-talk: a novel promoter of tumor progression
.
Cancer Res
2015
;
75
:
1777
81
.
39.
Neman
J
,
Termini
J
,
Wilczynski
S
,
Vaidehi
N
,
Choy
C
,
Kowolik
CM
, et al
Human breast cancer metastases to the brain display GABAergic properties in the neural niche
.
Proc Natl Acad Sci U S A
2014
;
111
:
984
9
.
40.
Minami
K
,
Liu
S
,
Liu
Y
,
Chen
A
,
Wan
Q
,
Na
S
, et al
Inhibitory effects of dopamine receptor D1 agonist on mammary tumor and bone metastasis
.
Sci Rep
2017
;
7
:
45686
.
41.
Wang
S
,
Mou
Z
,
Ma
Y
,
Li
J
,
Li
J
,
Ji
X
, et al
Dopamine enhances the response of sunitinib in the treatment of drug-resistant breast cancer: involvement of eradicating cancer stem-like cells
.
Biochem Pharmacol
2015
;
95
:
98
109
.
42.
Peverelli
E
,
Giardino
E
,
Treppiedi
D
,
Locatelli
M
,
Vaira
V
,
Ferrero
S
, et al
Dopamine receptor type 2 (DRD2) inhibits migration and invasion of human tumorous pituitary cells through ROCK-mediated cofilin inactivation
.
Cancer Lett
2016
;
381
:
279
86
.
43.
Hoeppner
LH
,
Wang
Y
,
Sharma
A
,
Javeed
N
,
Van Keulen
VP
,
Wang
E
, et al
Dopamine D2 receptor agonists inhibit lung cancer progression by reducing angiogenesis and tumor infiltrating myeloid derived suppressor cells
.
Mol Oncol
2015
;
9
:
270
81
.
44.
Roy
S
,
Lu
K
,
Nayak
MK
,
Bhuniya
A
,
Ghosh
T
,
Kundu
S
, et al
Activation of D2 dopamine receptors in CD133+ve cancer stem cells in non-small cell lung carcinoma inhibits proliferation, clonogenic ability, and invasiveness of these cells
.
J Biol Chem
2017
;
292
:
435
45
.
45.
Feng
DD
,
Cai
W
,
Chen
X
. 
The associations between Parkinson's disease and cancer: the plot thickens
.
Transl Neurodegener
2015
;
4
:
20
.
46.
Bajaj
A
,
Driver
JA
,
Schernhammer
ES
. 
Parkinson's disease and cancer risk: a systematic review and meta-analysis
.
Cancer Causes Control
2010
;
21
:
697
707
.
47.
Webster
WS
,
Leibovich
BC
. 
Exisulind in the treatment of prostate cancer
.
Expert Rev Anticancer Ther
2005
;
5
:
957
62
.
48.
Tomoda
R
,
Seto
M
,
Hioki
Y
,
Sonoda
J
,
Matsumine
A
,
Kusuzaki
K
, et al
Low-dose methotrexate inhibits lung metastasis and lengthens survival in rat osteosarcoma
.
Clin Exp Metastasis
2005
;
22
:
559
64
.
49.
Fujita
K
,
Kubota
Y
,
Ishida
H
,
Sasaki
Y
. 
Irinotecan, a key chemotherapeutic drug for metastatic colorectal cancer
.
World J Gastroenterol
2015
;
21
:
12234
48
.
50.
Boyle
A
,
Ondo
W
. 
Role of apomorphine in the treatment of Parkinson's disease
.
CNS Drugs
2015
;
29
:
83
9
.
51.
Jung
YS
,
Lee
SO
. 
Apomorphine suppresses TNF-alpha-induced MMP-9 expression and cell invasion through inhibition of ERK/AP-1 signaling pathway in MCF-7 cells
.
Biochem Biophys Res Commun
2017
;
487
:
903
9
.
52.
Zhang
X
,
Ye
YL
,
Zhu
H
,
Sun
SN
,
Zheng
J
,
Fan
HH
, et al
Selenotranscriptomic analyses identify signature selenoproteins in brain regions in a mouse model of Parkinson's disease
.
PLoS One
2016
;
11
:
e0163372
.
53.
Rosok
O
,
Pedeutour
F
,
Ree
AH
,
Aasheim
HC
. 
Identification and characterization of TESK2, a novel member of the LIMK/TESK family of protein kinases, predominantly expressed in testis
.
Genomics
1999
;
61
:
44
54
.
54.
Farkhondeh
A
,
Niwa
S
,
Takei
Y
,
Hirokawa
N
. 
Characterizing KIF16B in neurons reveals a novel intramolecular "stalk inhibition" mechanism that regulates its capacity to potentiate the selective somatodendritic localization of early endosomes
.
J Neurosci
2015
;
35
:
5067
86
.
55.
Hirokawa
N
,
Niwa
S
,
Tanaka
Y
. 
Molecular motors in neurons: transport mechanisms and roles in brain function, development, and disease
.
Neuron
2010
;
68
:
610
38
.