Glass tumor

Today's new cancer drugs that are customized to patients and their specific tumors are enabling sufferers to survive much longer. However, many cancers are resistant even to targeted drugs. Researchers at the University of Stuttgart’s Stuttgart Research Center Systems Biology (SRCSB) are currently analyzing tumor cells as a whole, in all their complexity, and using their findings to predict the efficacy of such drugs, both to develop new treatments and to be able to develop potential new drug candidates more rapidly.

It has long been known that tumors are as different as the people they affect. Even the cancer cells inside an individual person can differ, which is why chemotherapy, an undifferentiated chemical cosh aimed at all cancer cells, is increasingly being replaced by cancer drugs that target and attack individual changes within the cancer cells themselves, which promote the growth of tumors. Yet, even such personalized therapies often only show short-term results or none at all.

To ensure the success of a customized treatment, it is usually not enough to search for specific biomarkers in patients, such as gene mutations within their tumors or certain tumor-relevant proteins that may be being produced in greater or lesser quantities. The biology of tumors is far too complex for this. “That's like dismantling a radio and spreading the components across the table” as Morrison, Head of the Institute of Cell Biology and Immunology (IZI), explains: “the system will only function if I connect the heap of parts in a specific way”.

In future, it will be about not simply considering individual pieces of the puzzle, but rather trying to understand the tumor in its entirety, to develop drugs with more targeted effects. This discipline is known as system biology. Morrison’s research group, which predominantly includes cell biologists and systems biologists, is currently looking into the complex intracellular signaling paths, which seal the cell's fate, i.e., they either tell it to divide and multiply or self-destruct. In cancer, these signaling paths are often impaired and the cells start to divide uncontrollably, whereby the tumor cells usually exhibit multiple mutations in different, but often interconnected, signaling paths. Therefore, simply treating a single signaling path with drugs is unlikely to halt the growth of the tumor.

Usually, each cell continuously scans its environment via its receptors, and forwards information about external stimuli, such as growth stimuli or stress signals, via a series of messenger molecules within the cell. At the same time, the information flow branches out, is reinforced or incorporates data from other signaling paths. If the sum of all signals reach a certain threshold value, the information flows culminate in an order to the cell either to cease growing or to initiate the programed cell death process. It is easy to lose track of things among all the signaling noise, which is why Morrison's group is feeding high-performance computers with enormous amounts of data. They are linking data relating to signal messenger substance volumes from their own experiments on individual cells and animals to information about how the various messenger substances are interconnected and adding clinical data from cancer patients and their tumor properties.

From this they can produce circuit diagrams – like those produced for radios – for cancer cells and then run the signaling network and consequent changes on the computer. “Based on the mathematical models, we then try to predict, whether certain cancer cell lineages or tumors will be amenable to treatment with a given drug or what the best possible strategy is for making the tumor more sensitive again” says Morrison. The researchers’ predictions about colorectal and skin cancer cells have an 80 to 85 per cent probability of being correct, which, says Morrison, is already pretty good. His team then carries out empirical testing on the simulation so that they are able to configure the models with increasing precision. In future, the predictions could help protect patients from undergoing ineffective treatments and suffering their side effects. But, the virtual signaling network can also be used to develop novel drugs more cost-effectively and faster, as their effects on the signaling networks of tumor cells can be simulated on the computer. If one wanted to analyze all possible interventions in the network or modifications to drugs purely through empirical experimentation, it would take an inordinate amount of time and the task would be practically intractable.

The “Biomedical Development” and “Cell Biology” groups headed up by Roland Kontermann and Morrison’s predecessor at the institute Klaus Pfizenmaier respectively have already developed one very promising drug candidate together. It consists of a fusion protein, which drives cells to commit suicide, known as apoptosis, in targeted manner. The biologists modeled their creation on the TRAIL (TNF-related apoptosis inducing ligand) protein, which is produced by certain immune cells and triggers apoptosis in cancer cells, but has a negligible effect on normal healthy cells. “Tumor cells”, as Kontermann explains, “are already teetering on the brink; they just need a little shove.

You need to put a bit more effort in when it comes to healthy cells, as they can protect themselves much better”. However, clinical studies on the natural suicide messenger substances have so far delivered disappointing results. To trigger the cell death mechanisms in cancer cells more effectively, the protein engineers in Kontermann's team first used genetic engineering to link the three subcomponents of the TRAIL protein to form a single molecular chain. They then fused two such chains together, such that the designer protein they created can activate six death receptors on the cancer cells simultaneously.

Finally, they appended another antibody fragment, which specifically binds to certain surface molecules on tumor cells thus guiding the active substance to its target with a high degree of accuracy. “It looks very promising when used in colorectal and skin cancer cell cultures”, says Kontermann: “the TRAIL fusion protein also shows beneficial activity in mouse tumor models”. In the meantime, he goes on, three companies have indicated an interest in further developing the active substance in clinical studies. Meanwhile, the researchers working with Kontermann and Morrison are trying to completely destabilize tumor cells and give them the coup de grâce through a combination of the TRAIL fusion protein and chemo therapy drugs as well as newer substances. In the meantime, Morrison's team is augmenting its prediction model so that it can be applied to the combination therapy.

Whether genes are read and transcribed or not depends on small groups of chemicals on the DNA or DNA packaging proteins in the cell's nucleus, which are placed there or removed by enzymes. If enzymes attached methyl groups to the DNA base Cytosine at the starting area of a gene, they usually block the path of the gene transcription mechanisms. On the other hand, methyl, acetyl, phosphate or ubiquitin groups appended to the packaging proteins, the histones, cause the two-meter long DNA thread to wrap itself around a single histone complex, sometimes tighter, sometimes less tight, like a string of pearls.

Genes in densely packed DNA areas are also not transcribed and are virtually switched to silent mode. Researchers have discovered over 60 epigenetic markers thus far. In addition, there is also the fact that certain modifications usually interact in groups, which influence one another reciprocally. “We have 100 times more epigenetic than genetic information, because the human body comprises around 200 cell types, which carry different epigenetic markers” says Jeltsch. One of the objectives of a biochemist is to create a bit of order in this jungle of modifications and to understand which combinations of epigenetic markers work in which ways and may potentially trigger cancer.

Jeltsch’s group has taken a first system-biological step in this direction by developing a new tool for detecting two adjacent epigenetic markers on histones at the same time, which they have since patented. The trick: Jeltsch's team fused two different epigenetic protein transcription domains, which specifically attach to histones if they are carrying certain markers. In theory, the researchers could combine any two different transcription domains, which would enable them to analyze every possible pair combination of histone modifications across the entire genome.

To date, researchers have had to search for pairs of epigenetic markers in two sequential steps using specific antibodies. “We're faster with our one-step process and need less source material”, Jeltsch says. Moreover, the properties of antibodies could vary from batch to batch, so that the experiments are not always repeatable. Another of Jeltsch’s research focuses are methyltransferases, attach methyl groups to DNA or histones. Mutations in these epigenetic enzymes have been discovered in some cancer patients. Jeltsch and his team are interested in how these mutations affect the function of methyltransferases. “Ideally, if we understand what is happening there”, Jeltsch continues, “we could adjust the therapeutic treatment of patients, who have these mutations accordingly”. Methyltransferase inhibitors could be useful against some mutations, for example.

That would be personalized medicine in the true sense. The researchers recently discovered, for instance, that the most common mutation in the DNA methyltransferase DNMT3A results in an altered DNA methylation pattern in a certain type of leukemia. Because methyltransferases unfold their effects across the entire genome, an altered DNA methylation pattern could affect many cancer-relevant genes. In the experiment, the researchers first analyzed just 56 different DNA locations. The goal now is to find out which methylation sites are marked incorrectly throughout the genome, and which genes are affected by this. “The individual pieces of experimental data then need to be recombined”, Jeltsch explains: “to do this, we need the help of our colleagues in the system sciences, who view the whole thing as a network to discover how all these modifications are interconnected and result in diseases such as cancer”. Whoever has an insight into the tumor cell network and knows which adjusting screw they can twist, will also be better placed to discover the tumor's Achilles heel.

The regular meetings at the Center, such as the “System Biology” seminar series, conferences, but also internships and workshops on the topic for up-and-coming scientists are all conducive to the desired exchange. “One cannot help but become acquainted with the contents of other disciplines” Morrison finds, “which, of course, expands one's own horizons”. Jeltsch and Morrison both agree that many interdisciplinary projects would never have come about without the SRCSB. On the one hand, the bio-scientists would never even have known what the system scientists, for example, were researching at the university at any given moment. On the other hand, it is an advantage when applying for third-party funding. “One would never expect to find expertise in mathematical models for predicting therapeutic outcomes at an institute of cellular biology,” says Morrison: “however, the SRCSB can convince an expert evaluator that we do have the expertise in house”. In the current five-year funding period, which finishes at the end of the year, the members were able to attract funding for 60 projects in the fields of system and synthetic biology.

Jeltsch and Morrison's research has also benefited from the collection of microscopes and large-scale image analysis equipment, which the institutes contribute and which are centrally managed and maintain by an SRCSB appointee, who also advises and supports the researchers with their measurements. “Normally, there is no one who exclusively deals with these things and trains people properly in their use”, Morrison says. In the coming funding period,the SRCSB intends to establish a similar technology platform to carry out system-biological research into proteins and metabolic byproducts. In previous years, the biosciences at the University of Stuttgart have undergone an increasing change away from traditional subject areas toward system- oriented research approaches, which is evident just from the new professorships in such subjects as computational biology and system biology. Morrison is convinced: “if we want to focus on system biology and establish an international reputation in this field, then that is only possible in the context of a structure such as the SRCSB”. Cancer research could also benefit from this.

Helmine Braitmaier