In this project, we wished to know how functionally distinct stem, progenitor, and more differentiated cell types within the malignant medulloblastoma clone transitioned from one cell state to another, and whether disruption of these cell fate transitions could have therapeutic potential.

Figure 1. Visualisation of the lineage hierarchy in SHH-MB.

Framing the Question: What Governs Tumour Emergence?

Our starting point was a foundation of knowledge built on years of research by our lab and others. It is now well-established, at least in the mouse, that Sonic Hedgehog subtype-medulloblastoma (SHH-MB) arises from a disrupted granule cell lineage hierarchy in the cerebellum, with stem-like cells at the apex (Fig. 1).^1,2 These cells, previously identified as key players in normal brain development, persist abnormally in the developing cerebellum during tumour formation in mouse models of SHH-MB.³ This latter finding suggests that failure of elimination of cells with potent developmental potential, even when quiescent, can lead to medulloblastoma formation through their execution of an aberrant developmental program by making a large number of proliferating and differentiating progeny.

Working closely with Prof. Ben Simons at Cambridge on data generated in our previous study,¹ we used mathematical modelling to deduce the growth rates of various functionally distinct cell populations found within SHH-MB tumours, including the rare, slowly cycling and potently tumourigenic stem cells. Pairing this analysis with single-cell RNA sequencing data, we see that these neoplastic stem cells exist in two states: quiescent and proliferative. It is the latter state that gives rise to the progenitors that rapidly divide and form the bulk of the tumour. One crucial question arose: what governs the transition from a quiescent stem cell to a proliferatively activated one, initiating the cascade that leads to tumour expansion?

Dissecting Quiescence and Activation: A Deep Dive into Cell Fate

To investigate what drives stem cells from quiescence to activation, we employed whole-genome approaches in mouse SHH-MB tumours in which the relevant cellular markers were tagged with distinct fluorescent proteins. The GFP⁺ SOX2⁺ stem cell population was sorted into proliferative RFP⁺ (MKI67⁺) and quiescent RFP^– (Mki67^–) subpopulations for downstream analyses (Fig. 2).

Figure 2. Experimental set-up to sort quiescent and activated stem cells for downstream assays.

These tools allowed us to perform high-resolution profiling of the transcriptional and chromatin accessibility signatures in these proliferatively distinct cell states. Defining the differences in these closely related populations raised the possibility of targeting a crucial stem cell fate transition during the process of tumorigenesis.

Among the many genes examined, Olig2 stood out. Its role in normal neural stem cell (NSC) proliferation and fate specification, and expression in glial tumours were better understood,⁴ but we now suspected it might be driving the corrupt NSC activation and tumour progression in SHH-MB. Collaborating with Drs. Xi Huang and Siyi Wanggou, we identified that this role of Olig2 extended across models of SHH-MB, validating and broadening the scope of our findings. These discoveries set the stage for a series of targeted experiments.

Decoding Olig2’s Role: From Big Data to Targeted Experiments

Uncovering Olig2’s potential involvement in stem cell fate transitions was an exciting breakthrough, but proving it required rigorous functional experimentation, particularly as its expression was confined to the very small fraction of tumour stem cells. We utilised CRISPR-mediated gene editing to evaluate its role in stem cell function, as well as their ability to form tumours in mice. The initial results were striking—Olig2 was not just involved but appeared to be a critical driver of stem cell activation and tumorigenesis, not by driving every tumour cell, but by precisely targeting the cell fate event that occurs to activate the quiescent stem cells.

As we pondered the implications of our findings, an unexpected opportunity arose: a collaboration with Curtana, a pharmaceutical company that was developing a small-molecule inhibitor (CT-179) targeting OLIG2 protein for glioma treatment. The inhibitor works by interfering with OLIG2’s transcriptional function. They freely provided us with the compound with no strings attached, which allowed us to directly test OLIG2 inhibition in our SHH-MB models. This collaboration was transformative, allowing us to pursue mechanisms of tumour development more deeply, and shifting our project from basic science into research with potential clinical applications.

Proof of Principle: Targeting Tumour Emergence at its Source

Despite constituting only 1–5% of end-stage tumours, MB-stem cells play a disproportionate role in tumorigenesis. Conventional treatments of SHH-MB tumours target the rapidly cycling tumor bulk, which are the downstream output of the stem cells. However, the tumour initiating stem cells themselves are spared, and can regrow the tumour to cause relapse.^1,5

Inhibiting OLIG2 in our SHH-MB mouse model yielded striking results. By targeting OLIG2 in mice prone to SHH-MB during pivotal phases of tumorigenesis (tumour initiation or immediately following anti-proliferative “debulking” therapy), we effectively disrupted the transition from quiescent stem cells to proliferative stem cells and proliferative progenitors, constraining or “caging” stem cells in their quiescent state, and halting tumour progression (Fig. 3). This proof-of-concept demonstrated that stem cell fate transitions are essential, not incidental, to tumorigenesis in medulloblastoma, and perhaps other tumour types as well. Interestingly, treatment of fully formed tumours alone is insufficient to treat the tumour and prolong survival of mice because the large proliferative population drive tumour mass expansion and kills the animals. This suggests that targeting stem cell activation during critical windows are key to mitigate the uncontrolled proliferation of downstream tumour progenitors.

Figure 3. Model depicting OLIG2 inhibition mitigating SHH-MB primary or relapse tumour development in a mouse model; all figures created in BioRender.com.

The implications are potentially profound: rather than targeting only fast-dividing tumour cells and leaving behind stem cells that could drive relapse, a complementary strategy constraining the tumour-driving stem cells during critical windows of tumor development could potentially enhance treatment outcomes. Interestingly, this inhibitor also showed efficacy in DIPG models, suggesting applications beyond childhood medulloblastoma for tumour cells that express Olig2. 

Future Directions: Toward Precision Therapies for Medulloblastoma

Our experimental findings highlight the power of studying tumour development from its earliest stages, rather than studying tumours only at endpoint. This approach allowed us to dissect the mechanisms of medulloblastoma formation, and to demonstrate that targeting a rare cell state transition has important implications in mitigating tumour growth. Perhaps in the future, such a strategy could be used to intercept tumour formation– if early lesions are identified, particularly in people deemed at increased risk, treatment could be initiated, representing a new form of precision preventative medicine in neuro-oncology.

We have harnessed the power of the mouse model to understand and target the earliest steps of tumorigenesis. The next steps will involve refining OLIG2 inhibition strategies in the human context. Serendipitously, after making many of the key findings in our paper, we became aware of an independent international collaborative effort between two groups led by Drs. Tim Gershon (Emory) and Bryan Day (Brisbane), on OLIG2 inhibition in a range of preclinical models of SHH-MB, using the same inhibitor from Curtana.⁶ It was striking how complementary our findings were and sharing our data allowed both teams to perform additional experiments in revision that strengthened our conclusions, culminating in back-to-back publications, and we’d  also like to draw attention to their study. We are all excited to move CT-179 forward for clinical studies.

We are also enthused by our findings linking heterogeneous cell types within the malignant clone. Beyond a description of cellular heterogeneity, such as is being described by the incredible technology of single cell sequencing, we define a mechanism of stem cell state change that has implications for function of the cell types. Further identification of these cell fate decisions in brain cancer, in our view, will have implications for designing new therapies more generally for diseases characterised by corrupted developmental lineages.

In addition to Professor Ben Simons and Dr. Xi Huang who collaborated with us on this study, we also thank Drs. Gregory Stein and Santosh Kesari at Curtana Pharmaceuticals for generously providing us with the OLIG2 inhibitor. 

References:

1. Vanner, R. J. et al. Quiescent sox2(+) cells drive hierarchical growth and relapse in sonic hedgehog subgroup medulloblastoma. Cancer Cell 26, 33–47 (2014).
2. Schüller, U. et al. Acquisition of granule neuron precursor identity is a critical determinant of progenitor cell competence to form Shh-induced medulloblastoma. Cancer Cell 14, 123–134 (2008).
3. Selvadurai, H. J. et al. Medulloblastoma Arises from the Persistence of a Rare and Transient Sox2 Granule Neuron Precursor. Cell Rep 31, 107511 (2020).
4. Meijer, D. H. et al. Separated at birth? The functional and molecular divergence of OLIG1 and OLIG2. Nat Rev Neurosci 13, 819–831 (2012).
5. Zhang, L. et al. Single-Cell Transcriptomics in Medulloblastoma Reveals Tumor-Initiating Progenitors and Oncogenic Cascades during Tumorigenesis and Relapse. Cancer Cell 36, 302–318.e7 (2019).
6. Li, Y. et al. Suppressing recurrence in Sonic Hedgehog subgroup medulloblastoma using the OLIG2 inhibitor CT-179. Nature Communications 16, 1–23 (2025).