Metastasis, as a Cause of Cancer Related Mortality

More than 90% of of cancer related mortality is caused by metastasis. To develop new therapeutic strategies it is vital to understand the initiation and progression of metastasis. To identify and isolate metastasis initiating tumour-cells scientists developed a fluorescence-activated cell sorting (FACS)-based array. There are two types of metastasis-cells: metastasis-cells from low-burden cells and metastasis-cells from high-burden cells. After being transplanted, low-burden metastasis-cells showed that they have a considerable amount of tumour initiating ability and could differentiate to produce luminal-like cancer cells. When low-burden metastatic cells progress to high-burden metastasis cells they get an increased proliferation and MYC expression. This can be weakened with the use of inhibitors. This all supports a hierarchical model in which metastasis is initiated by stem-like cells, and can progress from low-burden to high-burden metastasis cells. The human breast contains two kinds of epithelial lineages: the basal/myoepithelial which contains stem-cells, and the luminal lineage which contains progenitor cells and mature cells.

The scientists used breast tissue from the mammoplasty of three individuals. With the use of numerous statistical analysis they concluded that basal/myoepithelial and luminal lineage is different for everyone.They focused in this experiment on a particular subtype, because this subtype is the most aggressive and there is no suitable treatment for it. The patient-derived xenograft maintained the same properties in the mice as in the patients and because of this it was suitable for the studies of human metastasis. To isolate the metastasis cells from the patient derived xenograft mice they developed a new FACS based array. With this they were able to detect metastatic cells in 70% of the patient-derived xenograft peripheral tissue from the mice. The mice were analyzed when their tumours reached 20-25 mm in diameter. The growth kinetics was consistent within each model. Although the tumour of every animal had about the same diameter they all had great variation in metastatic burden.

The scientists also found out that PCA plots for mice with low burden metastasis cells were further away from the tumour they derived from than high burden metastasis cells. Other experiments also showed that low burden metastasis cells liked to form clusters with each other while high-burden metastasis cells liked to form clusters with primary tumour cells. The scientists found out that low burden metastasis conserved their basal/myoepithelial signatures. They had expressed higher levels of 22 basal/myoepithelial genes and expressed lower levels of 7 luminal genes. By focusing on clustering only the metastasis cells the scientists discovered incredible heterogeneity in differentiation, which correlated with metastatic burden. Akin to the mammary gland metastatic cells organized into two different clusters, where the low-burden cells were the most basal/stem-like and the high-burden cells were the most luminal-like. The scientists concluded the same conclusion with lung metastatic cells. Which means that it is a conserved phenomenon in each model. There were some differences between gene expression of lung metastatic cells of different models, but they were not enough to cluster metastatic cells separately by patient derived xenograft models.

In order to investigate heterogeneity at protein level scientists performed immunostaining for a basal and luminal gene. Tumour cells found in micrometastatic from low-burden tissues had a high percentage for the basal gene and luminal gene and tissues from high-burden tissues had a high percentage for the luminal gene and were heterogeneous for the basal gene. This suggest that differentiation status correlates with metastatic burden in protein-level.By means of single cell analyses scientists discovered that in the low-burden metastatic cells had high levels of pluripotency genes. These genes suggest that they are exploited embryonic programs for self-renewal and maintenance. Low-burden metastatic cells also expressed higher levels of typical EMT-markers, except for an EMT-marker which was typically found in normal basal/stem cells. All these findings are consistent with previous reports which show that EMT promotes stemness in mammary gland, and suggest that low-burden metastatic cells utilize an EMT-program to make dissemination easier.

Further studies also revealed that genes involved in the DNA damage response, chromatin modification, differentiation, apoptosis and the cell cycle were differentially expressed in low-burden metastatic cells.Because of the heterogeneity in metastatic cells scientist wondered if stem-like cells directly give rise to luminal-like cells, or if the luminal cells are originated from founder-cells. After an experiment the scientist concluded that luminal-like cells can derive from cells that disseminate at the early stages of primary tumour growth. To test the growth and differentiation capacity of stem like metastatic cells, scientists transplanted low-burden metastatic cells into mammary glands. Interestingly 2 of 4 transplanted cells produced large tumour, while primary tumour cells never produced tumours, even at 100 fold high numbers. This is consistent with the previous reports which showed that PDX tumours are more efficiently increased as fragments than dissociated cells.

After single cell analyses scientists concluded that low burden-metastatic cells have high tumour initiating capabilities, and that they can give rise to luminal-like tumour cells. This supports the hypothesis that stem-like metastatic cells give rise to luminal metastatic cells. Another interesting question the scientists had was is stem-like cells were present in tumour cells, or if they evolve after interaction with their microenvironment. After a test the scientists concluded that primary tumours contain a rare subpopulation of stem-like cells, and that the percentage correlates with metastatic potential. Afterwards scientists wanted to know if enrichment of this stem-like signature in primary tumours may be predictive of distant metastasis in human patient data sets. After an analyses the scientists found that 16 of 55 genes associated with stem-like metastatic cells were significantly prognostic. Previous work has shown that metastatic cells in different organs display specific gene expression signatures. Supervised clustering by the targeted organ has shown that metastatic cells in brain, bone marrow and peripheral blood have differences in gene expression patterns. Brain metastatic cells are the most different.

CTC’s are very important for diagnosis. Most CTCs and bone marrow DTCs clustered with intermediate metastatic cells. This might have been because the cells were from animals with intermediate burden. However, 16.7% and 10.7%, showed a more basal/stem-like signature, which sugges that these stem-like cells may represent the initial metastatic seeder cells.Scientists also observed a shift towards a more quickly increased signature which have been correlated with increased metastatic burden. Low-burden metastatic cells expressed higher levels of rest and dormancy-associated genes. Higher-burden metastatic cells appeared to enter the cell cycle, expressing lower levels of quiescence and dormancy-associated genes and higher levels of cell-cycle-promoting genes. Scientists also discovered primary tumour cells (22.2%) with this less-proliferative signature.

These discoveries inspired scientists to test if blocking this switch from inactivity into the cell cycle could stop metastatic progression. Since scientists detected high levels of both MYC and CDK2 in more advanced stage metastatic cells , the scientists chose to test a CDK inhibitor that has been shown to end apoptosis in high MYC-expressing cancer cells. The scientists came up with a hypothesis that apoptosis would be started in metastatic cells progressing into proliferation, since they appear to upregulate MYC. After testing this on mice the scientists found that by looking in high resolution at gene expression in single metastatic cells, scientists have uncovered an previously unrealized range in differentiation and gene expression linking to the metastatic stage ,and demonstrate that this approach can ease the recognition of new potential drug targets with efficiency against metastatic disease.METHODSTo begin with the analysis the researchers first gathered the cell lines and the xenografts of the tumour tissues, which were grown and acquired according to standard and ethical protocols.

The xenografts were divided into tumour fragments and propagated into the breasts of the mice. When the tumours became palpable, that’s when the tumours were measured weekly to oversee their growth rate. The tumour fragments were stored by freezing them in liquid nitrogen. All animals from which xenografts were derived were euthanized at the end, when the tumours had grown to about 20 to 25 mm in size. During the resection experiment, tumours were usually removed when they reached the size of about 10 to 12 mm. The animals on which resection was performed were brought back to their colony and were warranted to grow metastases for 8 weeks, during which lung tissue was gathered and analysed by fluorescence-activated cell sorting (FACS) for human cells.In order to measure the functional activity of metastatic cells, orthotopic transplant experiments were performed on the animals. Particular metastatic cells in the lymph nodes, as well as particular tumour cells from matched animals, were segregated by FACS and combined from various animals. The sorted cells were formed into pellets and inserted into a media. Diluted versions of these were inserted into the breasts of 3.5-week-old mice and grafts were taken after 4.5 months when the primary tumours became 20 mm in size.

After this began the dinaciclib treatment experiments, which were administered when the tumours became palpable. The dinaciclib was primed and acquired according to protocol. The mice were randomly appointed to treatments when the tumour cells were transplanted and analysed with the help of the single-blind design. In total, 49 animals were injected with the treatment three times a week. Animals were measured twice a week to report primary tumour growth. The mice were euthanized at the end of the treatment or earlier if the tumour reached 20 mm in diameter. Animals which developed unfavourable effects were ruled out of the study. The microarray gene expression values were calculated using some form of statistics program. Plasma membrane genes expressed greatly on all of the 15 tumour sample xenografts. The 12 initial patient tumour samples were ranked from highest to lowest expression. The predicted value of every one of the 55 genes characteristics of low-burden metastatic cells was worked out by Kaplan–Meier analysis.

All solid tissues and the brain were dissociated for FACS. The tissues were cut up and placed in culture medium. They were then broken down for 45 min at 37 °C. The suspensions that arose were then inserted into a solution of DNase for 3 min at room temperature, after which they were washed and dissociated again. After this peripheral blood, supernatant and bone marrow were collected, cells were pelleted for 5 min and leftover erythrocytes in peripheral blood, lung and tumour samples were lysed for 5 min at room temperature. All unused samples were directly filtered and stored by freezing them in liquid nitrogen. The tissues from the reduction mammoplasty were washed three to five times, cut into small fragments and digested overnight in a solution. The digested fragments were then pelleted for 3 min, frozen and then stored in liquid nitrogen. The antibodies for several particular human antigens were bought commercially. Both human and mouse antibodies were stained. After 15 min of lying on ice, the stained cells were washed to get rid of excess antibodies and put back into the medium. The cells were then flow sorted and analysed. Dead cells were eliminated and contaminating human or mouse haematopoietic and endothelial cells were excluded. The complete tissue sample in the single-cell multiplex qPCR experiments was run through the flow cytometer. A steady number of live cells were found in the tissues of all of the animals.

The results of mice which deviated by more than one standard deviation were excluded from the study. Single-cell gene-expression experiments were carried out with microfluid chips. Single cells were sorted using FACS into distinctive wells. The experiments were done according to protocol. Each well was prefilled with a solution. After the sorting process, the PCR plated were frozen and or placed into the thermocycler to go through the process of combined reverse transcription and target-specific amplification. Exonuclease reaction solution was subsequently added to remove unincorporated primers. Each well was then diluted. A bit from each sample was then dropped into a separate plate and mixed with another solution. Individual primer assay mixes were made in yet another plate. The chips were primed before the samples and assays were mixed into them. The chips were then evaluated thoroughly. All of the single-cell PCR data were analysed using a statistical analysis software. In its entirety, 268 mammary cells from reduction mammoplasties as well as 441 metastatic and 523 primary tumour cells from the xenografts of the mice were analysed.

The results of the analyses were developed into Ct values, which were then further generated into statistical language.In regular mammary cell experiments, the Ct values were standardized by deducting the average value of the basal/stem-cell population per gene and per array. In the mice xenograft experiments, the Ct values were standardized by deducting the average primary tumour expression per gene as well as the average value of the basal/stem-cell population per gene and per array. Low quality samples were found and withdrawn from additional analysis.Various statistical tests were performed in order to determine gene expression differences between earlier established populations. For regular mammary cell experiments, a threefold comparison was initiated between basal/stem, luminal, and luminal progenitor cells. This generated an array of 49 differentially expressed genes. To find out of which population each gene is an aspect of, pair-wise tests were executed. When comparing metastatic cell experiments to that of primary tumour cells, only the pair-wise tests were executed.

Threefold comparisons were executed to compare lung metastatic cells from the three xenografts from the mice and fivefold comparisons were executed to compare metastatic cells from each tissue. These analyses were done with a variety of statistical programs. In order to find passages which were represented in a greater fashion in the set of significantly differentially expressed genes that they would have been by just chance, an enrichment analysis of Biological Process gene ontology terms was executed using several statistical programs. For both histological analysis and immunofluorescent analysis, the tissues were suspended overnight in paraformaldehyde and processed paraffin embedding. The tissues were stained with haematoxylin and eosin for histological analysis. In order to commence the immunofluorescent analysis, the tissues were stained using immunofluorescent staining. The immunofluorescent staining was used upon lung tissues with low and high metastatic burden.On the sections with paraffin-embedded tissue immunostaining was carried out by using a citrate buffer and heating the sections in a pressure cooker for 8 min. Several human genes were stained with the help of a three-step method. First, the primary antibodies were incubated overnight, which was then followed by one-hour incubations along with detecting antibodies after which a fluorescent binding biotin was added.

MYC and phospho-histones were found by following a two-step method, in which the overnight staining of antibodies was followed up by a one-hour incubation along with detecting antibodies. The number of positive nuclei was counted for tumour, high burden, and low burden cells, after which the significance was calculated by a statistical analysis software.