Not all tumor cells are created equal

Introduction & Background

Stefanie Jeffrey, MD (2)

Earlier in the quarter Dr. Islas taught us that three lines of evidence – biochemical, immunological, and cytogenetic – suggested tumors are monoclonal growths.  Recall in a monoclonal growth, a single cell transforms from normal to malignant and subsequently gives rise to a tumor composed of genetically identical cells.  Together, the three experiments, from different fields, provided strong support that cancer arises from a single progenitor cell gone awry.

Alternatively, a tumor can be polyclonal in origin. In this case, multiple cells transform from normal to malignant, leading to a tumor mass composed of genetically distinct sub-populations of cells. (1) (Fig. 1)

Over the weekend, I came across a study conducted by a group of Stanford scientist who have discovered that cancer cells shed by a single tumor into the bloodstream are genetically diverse.  Some cancer cells have turned on genes that make them more adept at lodging themselves in new places, aiding in their ability to metastasize to new organs (2).  Other cancer cells have an entirely different pattern of gene expression.
The senior author of the study is Stefanie Jeffrey, MD, professor of surgery and chief of surgical oncology research at the Stanford University School of Medicine.  The research was published in PLoS ONE on May 7, 2012.

What are Circulating Tumor Cells (CTCs)?

Circulating tumor cells (CTCs) are rare cells found in the blood of patients with solid tumors (3).  They are epithelial cells shed from the tumor itself, and are believed to play an important role in cancer metastasis.  Previous studies have found the number of CTCs in blood to be tightly linked with clinical outcome in patients with metastatic breast, prostate, colorectal, and lung cancer (4, 5, 6, 7).  Therefore, isolating CTCs and identifying their genetic makeup, may provide us with necessary information to one day be able to target these cells and slow cancer dissemination.

Since CTCs are accessible by an easy blood draw from the cancer patient, extracting and genotyping CTCs would seem to be a fairly easy task.  However, CTCs are present in the blood of cancer patient’s amidst 5x10⁹ (billions) red blood cells and 5x10⁶ (millions) white blood cells per ml (3).  Therefore, due to their rarity, separating CTCs from blood cells is extremely difficult.


In this study, Jeffrey and her team focused on studying CTCs from breast cancer patients.  Initially, the researchers collected blood samples from 50 participants: 20 primary breast cancer patients without detectable metastatic disease, and 30 metastatic breast cancer patients (3).  All individuals were consented patients of the Stanford Breast Oncology Clinic.

Isolating CTCs: MagSweeper

In order to overcome the challenge of extracting CTCs from blood, Jeffrey in conjunction with a team of engineers, quantitative biologists, genome scientists, and clinicians, developed a machine called the MagSweeper.  The MagSweeper uses an immunomagnetic separation technology to extract live CTCs with very high purity from the blood samples of patients with breast cancer (2).

To isolate CTCs, Jeffrey and her team capitalized on a cell-surface protein present on epithelial cancer cells, but absent on healthy blood cells (2).  The protein is called EpCAM (epithelial cell adhesion molecule).  First, the scientists treated each blood sample from their breast cancer patients with antibodies against human EpCAM.  Attached to the anti-EpCAM antibodies were small magnetic beads.

Jeffrey then had the MagSweeper device scan each blood sample twice. The MagSweeper is equipped with a magnetic rod, such that the movement of the MagSweeper during each scan produces a force that attracts any cell that has been bound by the anti-EpCAM-magnetic bead-antibody, while releasing other blood cells.  Captured cells were then released into a buffer for analysis later (3) (Fig. 2).

Is the MagSweeper specific for cancer cells?

           To test whether the MagSweeper was specific for epithelial cancer cells, the researchers ran a control.  Jeffrey used the device to scan blood samples from 45 patients without epithelial cancer: 25 healthy volunteers and 20 lymphoma patients.  No cells were captured, indicating that the MagSweeper was specific for epithelial cancer cells only (3).

Does the MagSweeper alter gene expression in cells?     

            To test whether the MagSweeper itself altered gene expression levels in cells, the researchers measured the expression of 15 genes in tumor cells derived from a primary breast cancer cell-line, called MC47.  Jeffrey compared gene expression before and after the magnetic bead-antibody labeling step.  She also compared gene expression before and after the MagSweeper capture process.  They found that expression of the 15 genes did not change during labeling or MagSweeper capture, indicating that the entire MagSweeper isolation process has no effect on gene expression (3) (Fig. 3A).

Does the MagSweeper affect cell viability?

To test whether the MagSweeper itself altered cell viability, the researchers measured plating efficiency of tumor cells derived from the same primary breast cancer cell-line used earlier, MC47.  Plating efficiency is the percentage of cells that produce viable colonies after seven days of growth.  Once again, Jeffrey compared plating efficiency before and after the magnetic bead-antibody labeling step, as well as plating efficiency before and after the MagSweeper capture process.  They found that plating efficiency did not change during labeling or MagSweeper capture, indicating that the complete MagSweeper isolation process has no effect on cell viability (3) (Fig. 3B).

Defining CTCs
            To narrow down the EpCAM-captured cells (510 cells total) from the blood samples of their breast cancer patients, Jeffrey developed a series of criteria for defining CTCs.

            First, the cells needed to express three reference genes: actin, beta (ACTB), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and ubiquitin B (UBB).  The ACTB, GAPDH, and UBB genes were selected because their expression is commonly seen in breast cancer cells.  For example, UBB gene expression has been validated in over 1,700 breast cancer samples (3).  After this initial screen, 63% (321/510) of the cells isolated by the MagSweeper qualified for further analysis (3).

            Next, the researchers ran the captured cells through another set of criteria.  Jeffrey and her team selected cells that had: 1) no expression of the white blood cell marker, CD45, 2) expression of any of the following epithelial cell markers – KRT7 (keratin 7), KRT8 (keratin 8), KRT18 (keratin 18), and/or KRT19 (keratin 19).  Recall that Dr. Jeffrey and her team initially collected blood samples from 50 participants: 20 primary breast cancer patients without detectable metastatic disease, and 30 metastatic breast cancer patients.  After this final screen, however, the scientists were left with blood samples from only 35 patients: 14 with primary breast cancer and 21 with metastatic breast cancer (3).

            Since some patients had more qualified-CTCs compared to others, the researchers randomly selected five CTCs per patient (3).

Gene Expression

            Once the researchers isolated their CTCs of interest, their next goal was to obtain the genetic profile of each individual cell.  Stephen Quake, PhD, Stanford engineer professor, invented real-time PCR microfluidic chips that allowed them to measure expression levels of 87 cancer-related genes along with the 3 reference genes (ACTB, GAPDH, and UBB) in individual CTCs at once (2).  In contrast to past studies where researchers analyzed CTCs in groups and took their average gene expression, Jeffrey wanted to look at circulating tumor cells one-by-one in order to detect differences between individual tumor cells.

            Jeffrey selected the 87 genes based on published literature.  They included genes common in molecular pathways relevant to breast cancer, breast cancer biomarkers, as well as genes associated with cancer signaling pathways, epithelial-mesenchymal transition, cancer stem cells, and metastasis (3).

Is the real-time PCR microfluidic chip accurate?

To test whether the real-time PCR microfluidic chips were accurate in detecting gene expression levels, the researchers ran a control.  Jeffrey and her team analyzed tumor cells from: three primary (CCdl054, CCdl672, CCdl675), and four metastatic (T47D, MCF7, SKBR3, MDA-MB-231) breast cancer cell-lines.  Used by breast cancer researchers and pharmaceutical scientist worldwide, the genetics of these seven tumor cell-lines are well-outlined, thereby allowing researchers to know if the chips work well (2).

First, the scientists extracted genetic material from individual cell-line derived tumor cells.  Next, Jeffrey used the PCR microfluidic chips to simultaneously measure expression of all 87 genes in each tumor cell.  Jeffrey and her team found that indeed, genes expressed in the tumor cells reflected the known properties of the cell-line models.  For example, 99% (48/49) of the cell-line CTCs expressed the human epidermal growth factor receptor 2 (HER2) and the epidermal growth factor (EGFR) – consistent with expected breast cancer biomarker patterns (3) (Fig. 4).  The results indicate that the PCR microfluidic chips are accurate at detecting gene expression.

Next, Jeffrey and her team extracted genetic material from individual CTCs derived from the blood samples of patients with primary and metastatic breast caner.  Once again, they used the PCR microfluidic chips to simultaneously measure expression of all 87 genes in each CTC.  The researchers found that 31 out of the 87 genes tested, were commonly expressed in 15% of the CTCs analyzed.  The 31 genes were associated with: 1) epithelial mesenchymal transition - TGFb-1, FOXC1, CXCR4, NFKB1, VIM, ZEB2; 2) metastasis – S100A9, NPTN, S1004A; 3) PI3K/AKT/mTOR pathway – AKT1, AKT2, PIK3R1, PTEN; 4) apoptosis – BAX, CASP3, CD53, CD59; 5) cell proliferation – RRM1, MAPK14; 6) DNA repair – PARP1; 7) cell metabolism – SLC2A1, TFRC; 8) stem cell phenotype – CD24, CD44 (3).

            Further analysis of the 31 most dominantly expressed genes showed two distinct groups of circulating tumor cells.  Cluster I was comprised of 21 CTCs from 13 patients, and Cluster II was comprised of 84 CTCs from 30 patients (3).  “Depending on which genes we used to divide the CTCs into groups, there were as many as five groups, each with different combinations of genes turned on and off,” Jeffrey explains (2) (Fig. 5).

            The results indicate that solid tumors contain a variety of genetically different cancer cells that may eventually get shed into the bloodstream.

            Applying this finding to clinical medicine means that a single biopsy from a patient’s tumor may not be sufficient nor representative of the entire population of cells within that tumor.  For example, one of the breast cancer patients in this study had some CTCs positive for HER2, and some CTCs negative for the marker.  When this particular patient was treated with a therapeutic used for HER2-positive cancers, the CTCs with HER-2 were eliminated while the CTCs lacking HER2 remained in her bloodstream (2).  Although for many reasons it is physically impossible to biopsy every single metastatic lesion a cancer patient may have, Jeffrey and her team believe CTC-blood draws may offer a non-invasive means of studying the wide variety of molecular changes that drive a cancer forward and help it to spread.

Gene expression in cell-line derived tumor cells vs. patient derived CTCs

            Lastly, Jeffrey and her team compared the genetic makeup of CTCs from breast cancer patients with tumor cells from the widely-used experimental cell-line models they studied earlier.  When the 31 most commonly expressed genes were considered, the researchers found that none of the human CTCs had the same gene patterns as any of the tumor cells from the cell-line models.  For example, CTCs from breast cancer patients had higher expression of genes encoding growth factors and their receptors compared to tumor cells from cell-lines.  Furthermore expression of genes encoding proteins that act as downstream effectors in cell cycle progression and proliferation, were also higher in CTCs from breast cancer patients compared to tumor cells from cell-lines (2) (Fig. 6).

            The results indicate that human derived CTCs have a different gene profile compared to cell-lines derived tumor cells.

            Applying this finding to the world of cancer research is a bit scary, because tumor cells from these cell-lines are what scientists worldwide are using to develop drugs as well as test drugs.  This means that drugs that appear to be effective in tumor cells from cell-lines may not be effective at all against human CTCs due to different genetic profiles.

Conclusion

            Ultimately, Jeffrey’s findings are the first to show the extent of the genetic differences between tumor cells.  By looking at tumor cells one-by-one, the scientists discovered that even in an individual patient, tumor cells that make it into blood vary drastically.

The fact that CTCs are thought to play an important role in tumor metastasis along with the fact that these cells are accessible by an easy blood draw, could be the key to tracking tumors and detecting early metastasis in patients.  The genetic variation found in individual tumor cells could also change the way tumors are treated – in that maybe a more diverse combination of chemotherapeutics to treat each cell type present may be required.  Lastly, the significant differences in gene expression found between tumor cells derived from cell-lines and CTCs derived from human patients could change the way cancer research, specifically in regards to therapeutic development, is conducted.

Question & Thoughts

After reading the paper, a few questions and thoughts came to mind:

1)   After reading the Hallmarks of Cancer: The Next Generation, the findings of Jeffrey’s team are really not surprising.  In their article, Hanahan and Weinberg outline genome instability and mutation as an enabling characteristic of tumor cells.  We know from Dr. Islas’ telomerase lecture that cancer cells actually want to acquire mutations, especially those that are advantageous for survival and growth.  This is evidenced by the phenomenon termed, “delayed telomerase activation,” in which cancer cells will actually wait until their genome has generated enough tumor-promoting mutations before activating expression of telomerase.  Hanahan and Weinberg attribute this enabling characteristic of cancer cells to the fact that “certain mutation genotypes confer selective advantage on subclones of cells, enabling their outgrowth and eventual dominance in a local tissue environment” (8).  Therefore, the genetic variance seen in each individual circulating tumor cell in this study could very well be a prime example of cancer cells generating genomic instability so that they are able to more easily metastasize and invade other areas of the body.

2)   Do cancer cells turn genes on and off at random?  Or do cancer cells select their genetic profiles, by knowing which genes will be most advantageous turned on or off?

Possible Answer: Although I do not know the exact answer to this question, it seems most likely that mutations will occur at random.  If you think about it, cancer cells only care about one thing – that is dividing, and diving fast.  Tumor cells are not concerned about how “pretty” that division is, so long as it happens fast.  Rapid division means careless DNA replication, which ultimately means increased chances of mutation during each cell cycle.  This can easily lead to a tumor being genetically diverse.  Therefore, once again, the genetic variance of each individual tumor cell that Jeffrey and her team saw really is not surprising. Overtime, we would expect natural selection to occur – selecting for cancer cells with the most “harmful” genetic profile, selecting for those cancer cells with the perfect combination of genes turned off and on.

3)   The real-time PCR microfluidic chip used to measure gene expression in this study is very similar to the gene expression array Dr. Islas discussed in class.  Researchers use these functional genomic tools to survey the expression levels of many genes in a tissue preparation.  Computerized analysis following the expression arrays then makes it possible to identify certain genes that may be over- or under-expressed compared to normal cell counterparts.  Similar to what was done by Jeffrey in this study, once a genetic profile of a tumor cell is known, researchers can then use this information to divide or stratify cancers into groups having certain biological properties.  Dr. Islas took this genetic profiling of cells a step further when he mentioned the use of personalized medicine.  This means that once the genetic profile of tumor cells, such as CTCs are known, clinicians may one day be able to tailor drugs and a design a treatment plan that is specific against an individual patient’s cancer.

4)   The previous question, question #3, leads to the following question: What happens if the gene profiles of CTCs constantly change?

If ever blood draws detecting CTCs actually becomes a clinical diagnostic tool, I suspect that physicians will have the tests repeated every so many months in order to make sure the patient has not developed any significantly new forms of CTCs.

5)   This may be useful information for groups interested in doing their cancer project on cancer metastasis.