Sven Nelander’s research on the systems biology of neural cancers
Invasion routes of glioblastoma and their patient-specific vulnerabilities
Each year over 200,000 people worldwide are affected by the brain tumor glioblastoma. The disease is notorious for its aggressive course and still lacks a treatment. A key difficulty lies in the tumour cells' ability to grow diffusely through the brain. Such diffusely growing cells, which occur in both nerve pathways and along blood vessels, are impossible to remove surgically. Therefore, it is important to understand exactly how the invasion takes place, which genes are behind it, and whether diffusely growing cells can be knocked out with drugs.
Until now, there has been a lack of effective ways to answer these questions. Our team has developed an integrated strategy based on a new biobank of 100 PDX models (patient-derived xenografts), established at our centre (Johansson et al, Cell Reports; unpublished data). The PDX models can be thought of as replicas of the patient tumours. This means that copies of the original tumour can be recreated and studied in a laboratory environment.
Our initial studies show that the tumour cells from different patients differ in terms of tumour growth. We have also shown that computational algorithms can connect such differences to both genes and drugs, which makes the biobank a very powerful and unique research tool.
Our team is pioneering the use of patient-derived cell biobanks as a tool to uncover growth and drug response in the brain tumour glioblastoma (Cell Reports 2020, NeuroOncology 2018, Science Translational Medicine 2015, EBioMedicine 2015,). In recent work, we used our in-house Uppsala-based biobank HGCC to report on pharmacological subtypes of GBM. We are currently using the HGCC models to uncover differences in brain tumour invasion, using several different approaches. A key goal is to uncover the genetic programs that mediate invasion along different anatomical routes, including single-cell /spatial methods, CRISPR, and computational modelling.
Data-driven drug discovery for pediatric neural cancers
Nervous system malignancies, such as medulloblastoma, diffuse midline glioma and neuroblastoma, cause a high proportion of childhood cancer deaths. Even though the underlying mutations are often known in these cancers, they still remain out-of-reach from traditional classes of oncology drugs.
To address this problem, our team is exploring how innovative analyses of big data sets can be used to discover new therapeutics. In one of our recent projects, we developed a new algorithm, TargetTranslator, which could identify new treatments for neuroblastoma (Almstedt et al, Nature Communications 2020). In a recent proof of concept study, we mapped the dynamic changes that occur in brain tumour cells (Larsson et al, Molecular Systems Biology 2020).
Presently, our team is exploring how data-driven methods can be used to target a broader range of pediatric cancers, including medulloblastoma and diffuse midline glioma. In a more theoretical line of work, we are also exploring how ideas from statistical physics can help us understand how tumor cells can be targeted. Looking ahead, this research can help meet a need for new therapeutics against these difficult cancers.
We are exploring how innovative analyses of big data sets can be used to discover new therapeutics. In one of our recent projects, we developed a new algorithm, TargetTranslator, which could identify new treatments for neuroblastoma (Almstedt et al, Nature Communications 2020). The picture shows a 3D microscopy scan of a Zebrafish embryo transplanted with tumour cells (green). The zebrafish model lets us test drugs for safety and efficacy, thereby reducing some of the need for mammalian testing.
Our team is investigating how tumour heterogeneity and drug response depends on time-dependent changes in individual tumour cells. In a recent study (Molecular Systems Biology, 2021), we have proposed a new strategy to construct models of cellular hierarchies from barcoded single cell experiments. The models, which we call State Transition and Growth models, are fitted from experimental data, obtained using barcoded single-cell sequencing and optical reporters. STAG models may have several potential uses in tumour biology and drug development.