Genomic Technology Reveals How Cancer Evolves Treatment Resistance

Introduction

Scientists have long known that when bacteria find themselves in a stressful situation, they can boost their genomes’ mutation rate. This sounds counter-intuitive—after all, mutations are usually bad news. But it turns out that mutations can actually give bacteria the flexibility to evolve and adapt in harmful environments. Now, in a study published this summer in Science¹, Dr. Arcadi Cipponi and his colleagues at the Garvan Institute reveal that bacteria aren’t the only ones that employ this strategy—human cancer cells do too.

In their study, Dr. Cipponi and his team detail how human cancer cells increase their mutation rates when exposed to anticancer drugs, and eventually acquire resistance to those drugs. The researchers used a variety of Illumina technologies to support their work, including targeted gene panels, whole genome sequencing, and transcriptomics. “Thanks to this technology, we were able to formally demonstrate this hypermutability,” says Cipponi, who is a senior research officer at the Kinghorn Cancer Centre at the Garvan Institute in Australia, “It would have been very difficult to show this with other approaches.” The team’s findings will help scientists better understand treatment resistance, and may also improve cancer therapeutics.

iCommunity spoke with Cipponi about his latest study in Science¹, how he and his team investigated adaptive evolution in cancer, the implications of their findings for disease treatment, and why genomic technology is now critical in cancer research and clinical applications.

Arcadi Cipponi, PhD is a senior research officer at the Kinghorn Cancer Centre at the Garvan Institute in Australia.

Q: What originally made you interested in cancer research?

Dr. Arcadi Cipponi (AC): My story goes back to when I was a university student, where I became fascinated by the complexity of this disease. After my degree, I began working on cancer research. I started with cancer immunotherapy—we tried to utilize human lymphocytes modified to express tumour proteins as a sort of vaccine to induce an anti-tumor immune response. What we saw is that most of the time, cancer cells were able to evade the immune response. They essentially become resistant to the immune system. So then I really became interested in the genomic and phenotypic plasticity of cancer cells. The next logical step was to understand the way they mutate and change continuously, and this is what drove my decision to move into the field of cancer genomics.

Q: How did you go about investigating treatment resistance for your Science¹ paper?

AC: What we initially wanted to know in our study was a very fundamental question: are there common mechanisms that human cancer cells utilize to adapt and evolve in response to pharmacologic pressures? Because we know that human cancers are genomically unstable and have very extensive genetic and phenotypic heterogeneity. So we knew that there must be something at the genetic level that had to be switched on or off.

For our study, we didn’t use anti-cancer agents like chemotherapies. We chose instead to use cytostatic agents that are not genotoxic, meaning that they don’t damage DNA. This decision gave us the opportunity to look at the endogenous mechanisms of mutagenesis. We took different cancers—we chose a well-differentiated liposarcoma cell line, two melanoma cell lines, and two breast cancer cell lines. The beauty of this approach is that every cancer type has different molecular targets, so we used different drugs directed against very different signaling molecules. We started by studying how the pharmacologic pressure affected the cells’ DNA damage. We asked: do we see increased levels of DNA damage in these cells? We measured this by confocal microscopy, using a very common phosphoproteomic assay. And we saw that every single cell line had an increased number of DNA lesions, which are potent inducers of mutations. They were progressively adapting and evolving in response to the drug.

"Different genomic technologies are necessary because you want to collect as much evidence as possible to have results that are reliable and convincing."

Q: What is stress-induced mutagenesis and how did you detect it?

AC: It is well known that bacterial populations exposed to stressful conditions, such as changes in pH or nutrient starvation, can rapidly increase their mutation rates. This ‘hypermutator state’ is achieved through repression of the mismatch repair pathway, which usually repairs single nucleotide mismatches. Bacteria also switch from using high-fidelity polymerases to repair DNA lesions to instead using the Y-family polymerases, which are error-prone. By doing all this, bacteria generate new genetic diversity. This ultimately gives them a better chance of acquiring advantageous mutations capable of fostering adaptive evolution.

We wondered whether this occurred in cancer cells as well. And we discovered that there is indeed repression of high-fidelity DNA repair pathways, similar to prokaryotic mechanisms. Our data combined with other evidence suggests that human cancer cells can dynamically regulate mutagenesis in response to stress, exactly as observed in bacteria. This is really an evolutionarily conserved mechanism.

In the Science¹ paper, we go into the underlying molecular mechanisms that likely regulate this adaptive mutagenesis. To do this, we performed a whole genome RNA interference screen. In this screen, we silenced every single gene of the genome—the idea was that if silencing a specific gene fostered adaptability, then this would have been indicative of a role in these processes of adaptive mutagenesis.

The class of genes that most interested us were the common genes—those that when silenced, can foster resistance to multiple pharmacologic pressures. We ended up focusing on the kinase mammalian target of rapamycin (mTOR) for several reasons. The first reason is that it’s a well-known regulator of the stress response. The second is that there’s a clear link between mTOR signaling and the DNA damage response. And third, mTOR has clinical implications. There are many agents targeting mTOR signaling used in medical oncology, but most of the clinical trials where mTOR inhibitors are used as a single agent give disappointing results, and our studies provide a possible explanation.

Q: What exactly is mTOR and how does it work?

AC: mTOR is an evolutionarily conserved stress sensor and a central regulator of cell growth and proliferation. Essentially it orchestrates the downstream signaling pathways to say to the cell, yes, the environment is good, or no, it is necessary to reduce metabolic activities and proliferation—there is nutrient scarcity, or there is an excessive accumulation of DNA damage. This is the reason why it was quite interesting. And what we found is that mTOR signaling is actually repressed during the early phase of the adaptive evolution, exactly when most of the genomic alterations are acquired.

Since mTOR signaling controls cell proliferation, the deregulation of this pathway is frequently observed in many types of cancers. For this reason, mTOR is an important therapeutic target. However, mTOR inhibitors have greater therapeutic effects on objective responses or progression-free survival than on overall survival. Essentially, this means that the inhibition of this pathway induces a short-term response where initially malignant cells stop growing, but during this stationary phase they evolve new adaptive mechanisms which cause the cancer to restart growing more aggressively. The way we explain this is that mTOR inhibitors trigger this hypermutability and increase the chance for cancer cells to pick up advantageous mutations. Clinically, this explains the limited efficacy of mTOR inhibitors used as single agents.

So the mTOR pathway is really what allows adaptive mutagenesis to occur in cancer cells?

AC: Yes, our data support this model. We’ve done extensive in vitro studies to validate our results from the genome-wide RNA interference screen. We clearly show that in the absence of pharmacologic pressures, the silencing of mTOR reduces the clonogenic potential of different cancer cell lines. This is exactly what we were expecting - you hit the brakes and cell proliferation slows down. But the interesting thing is that when the same cells were exposed to different pharmacologic pressures, they had an advantage over control cells, where mTOR was not silenced.

"Illumina has extensive bioinformatic tools to check the quality of the reads. It’s a very well established pipeline which is absolutely reliable"

Q: What were the different genomic technologies used in your study and how exactly they were used?

AC: We used two different strategies to sequence the genome. The first one is a targeted sequencing approach using Illumina’s HiSeq 2500 system—we had a panel of roughly 300 cancer-related genes. Although these data strongly suggested accelerated mutagenesis in response to the targeted anticancer agents, we decided to formally estimate the increase of mutation rates by exploring the impact of the pharmacologic pressures on a genome-wide scale.

What we did is really something extreme, because we sequenced several single cell-derived clonal populations. We wanted to see if during the clonal expansion, the genetic diversity generated by the resistant populations was significantly higher than the parental untreated clonal populations. Clones were sequenced to a very high depth (~120x), by using the Illumina NovaSeq 6000 platform. Because of this depth, we were able to identify most of the single nucleotide variants (SNVs), insertions and deletions, and copy number changes. These data revealed a significant increase of genomic alterations in the resistant cells.

Illumina’s sequencing technologies are truly state-of-the-art at the moment. The data are reliable, and thanks to this technology, we were able to formally demonstrate the hypermutability. It would have been very difficult to demonstrate this with other approaches. It could have been done, but it would have been less elegant and probably less convincing.

Q: You also used transcriptomics and shRNA screens for your research, correct?

AC: Yes, we used samples harvested at different time points during the adaptive evolution of the drug-resistant populations for transcriptional profiling. We used Illumina’s HiSeq 2500 system for that as well. It gave us important data about the expression changes of DNA repair pathways. It was extremely important for us to understand the impact of stress on the DNA damage response.

The genome-wide RNA interference screen is a different approach. Essentially, we used hairpins to silence a specific gene. By sequencing, you identify the hairpins that are enriched or depleted during the evolution of the population. This gives very clear data about the role of a gene in the mechanisms of adaptive mutagenesis. Again, the Illumina HiSeq 2500 platform allowed us to multiplex an enormous number of samples, sequence everything together, and come up with lots of data.

Q: Why is it necessary to use different genomic technologies when doing this kind of research?

AC: It’s necessary because you want to collect as much evidence as possible to have results that are reliable and convincing. We used targeted sequencing at the beginning to get a snapshot of the genomic alterations in these cells. The data were convincing, but not enough, so we had to switch to whole genome sequencing. Whole genome sequencing was needed to find out every single mutation acquired by the cells. Lastly, transcriptional profiling was necessary to confirm our hypothesis. It also gave us a lot of information about what was happening in the cells.

Q: How has your experience with Illumina sequencing data been?

AC: It’s absolutely great. Illumina has extensive bioinformatic tools to check the quality of the reads. It’s a very well established pipeline which is absolutely reliable. We never have any doubts about the quality of the sequencing. Many of our research projects and all clinical studies are entirely dependent on this technology. So, yes, these are great technologies.

Q: What are the next steps in your research?

AC: We are working on a synthetic lethal approach to exploit the vulnerabilities induced by the targeted anticancer agents. I don’t know whether this will have a significant clinical impact, but it is certainly worth trying to understand whether this strategy can enhance the therapeutic response to current anticancer agents. We are also working on some of the other adaptive mechanisms revealed by the whole genome interference screen. One of them is mediated by chaperones, which cancer cells are heavily dependent on for protein folding, due to the presence of extensive genomic alterations and environmental stress. The underlying mechanisms regulating the activity of chaperones in malignant cells are very interesting—there are clear implications for cancer progression and treatment.

Q: How has genomics changed the way you think about cancer research?

AC: Without these technologies, it would be impossible to study the mutational processes that drive progression of cancer cells. Next-generation sequencing technologies are also increasingly utilized in medical oncology to identify ‘druggable’ mutations – this is what we call ‘personalised cancer medicine’ and has truly revolutionized the way clinicians are treating cancer patients.

Learn more about the products and systems mentioned in this article:

Illumina DNA Prep
www.illumina.com/products/by-brand/nextera.html

Illumina RNA Prep
www.illumina.com/techniques/sequencing/NGS-library-prep/rna.html

Illumina NovaSeq
www.illumina.com/novaseq

References

1. Cipponi A, Goode D L, Bedo J, et al. mTOR signaling orchestrates stress-induced mutagenesis, facilitating adaptive evolution in cancer. Science 5 JUN 2020 6495: 1127-1131