Therapies that activate immune cells called T cells to target tumours are an efficient way to combat many types of cancer1. But an aggressive brain cancer called glioblastoma has proved a particular challenge for immunotherapies2. The blood–brain barrier protects the brain against immune-cell infiltration to prevent the potentially life-threatening effects of brain inflammation. This phenomenon is beneficial in normal circumstances, but it prevents T cells from reaching glioblastomas, making the tumours immunologically ‘cold’3. In a paper in Nature, Samaha and colleagues4 report a way to trigger infiltration of T cells into the brains of mice, thus making glioblastomas vulnerable to immunotherapy.
In the disease encephalitis, brain inflammation occurs because T cells that are typically excluded from the brain migrate across the blood–brain barrier. This migration is a coordinated process that requires activated T cells circulating in the bloodstream to adhere to endothelial cells, which line blood vessels. Adhesion is mediated by the binding of ligand molecules on T cells to cell-adhesion molecules such as ALCAM, ICAM-1 and VCAM-1 on endothelial cells5. These cell-adhesion molecules are expressed at higher than normal levels in encephalitis6. Binding between ALCAM and the T-cell ligand CD6 halts the progress of activated T cells through blood vessels, allowing subsequent binding by ICAM-1 and VCAM-1. Once binding of T cells by cell-adhesion molecules reaches a critical threshold, the T cells can migrate between endothelial cells and so out of the vessel into the brain.
In glioblastoma, however, the brain vasculature is reprogrammed such that endothelial cells produce little or no ICAM-1 and VCAM-17. If researchers could increase adhesion between T cells and endothelial cells in people with glioblastoma, as occurs in encephalitis, it might be possible to enable transendothelial migration of T cells.
Samaha and colleagues found that endothelial cells in glioblastoma overexpress ALCAM. They reasoned that, by engineering T cells to bind to ALCAM more firmly, they could enhance T-cell anchoring in the endothelium and subsequently improve transendothelial migration. To this end, the authors generated a synthetic ligand for ALCAM, derived from CD6. They engineered their molecule, which they named homing-system CD6 (HS–CD6) such that individual ligands interacted with one another to produce a multimeric protein. The researchers introduced the synthetic ligands into T cells using a retrovirus construct. They found that the presence of multimeric HS-CD6 on T cells enhanced adhesiveness between these cells and ALCAM-expressing endothelial cells and, as predicted, enabled transendothelial migration in in vitro models.
Samaha and colleagues also uncovered details of the molecular program by which HS–CD6 triggers transendothelial migration. On binding by ALCAM, HS–CD6 activates the protein SLP-76 in T cells. SLP-76 mobilizes the protein LFA-1, which moves to the cell surface and binds the few ICAM-1 molecules present on endothelial cells, further enhancing binding between T cells and endothelial cells. These changes also activate FAK, a protein that modulates the network of actin proteins that confer T-cell shape, enabling T cells to squeeze between endothelial cells, crossing the blood–brain barrier (Fig. 1).
The next step for the authors was to ensure that T cells entering the brain would home in on tumour cells. T cells harbour antigen-receptor proteins on their surfaces that bind to specific protein fragments called antigens on target cells, enabling T cells to pick out foreign cells for destruction. Samaha et al. engineered T cells to express an antigen receptor that was designed to bind to human epidermal growth factor receptor 2 (HER2) — an antigen produced by glioblastoma cells. They then introduced these cells into mice in whose brains human glioblastomas had been surgically implanted. T cells that expressed both HS–CD6 and the HER2-specific antigen receptor infiltrated the glioblastomas, leading to complete remission and long-term survival in most of the treated animals. By contrast, T cells harbouring only the antigen receptor (which are typically used for cancer immunotherapy) did not infiltrate the tumour.
This study lays out a viable strategy for immunotherapy in glioblastoma. But key challenges must be overcome before we can translate the discovery from mice to patients. For instance, ALCAM is expressed by a variety of cell types, including bone-marrow cells8. More studies will be required to assess whether the integrity and functions of these non-endothelial cells are affected by the approach. In addition, toxicity could be an issue if T-cell targeting damages healthy brain tissue, either directly or indirectly. Strategies to limit T-cell activation and lifespan using genetic ‘off’ switches have already been developed9, and could potentially be used to combat such toxicity. Encouragingly, the fact that the authors’ mice survived long-term suggests that the treatment did not cause severe toxicity in the animals. However, the group did not investigate the persistence and activity of the HER2-targeting T cells in the body, or examine whether the cells targeted non-glioblastoma cell types.
Finally, targeting T cells to brain tumours is only a first, albeit crucial, step in triggering an effective antitumour T-cell response against glioblastoma. T cells entering a glioblastoma will encounter a profoundly immunosuppressive microenvironment created by low oxygen and pH levels and immunosuppressive molecules. This did not harm T cells in the glioblastoma-harbouring mice, but these animals do not mimic many key features of true human glioblastoma — and immunosuppression will certainly pose a challenge in humans. A successful immunotherapy strategy for glioblastoma will ultimately consist of a combinatorial therapy that allows enough active tumour-specific T cells to enter and persist in an immune-permissive tumour microenvironment10. Such an approach would transform this deadly disease from an immunologically cold target to a hot one.
Nature 561, 319-320 (2018)