Tumour tamed by transfer of one T cell

The T cells of the immune system can be engineered to target a tumour, but why some people respond better than others to such therapy is unclear. One patient’s striking response to treatment now offers some clues.
Marcela V. Maus is in the Department of Medicine, Division of Hematology and Oncology, Harvard Medical School, and the Massachusetts General Hospital Cancer Center, Boston, Massachusetts 02114, USA.

Search for this author in:

The use of genetically engineered immune cells to target tumours is one of the most exciting current developments in cancer treatment. In this approach, T cells are taken from a patient and modified in vitro by inserting an engineered version of a gene that encodes a receptor protein. The receptor, known as a chimaeric antigen receptors (CAR), directs the engineered cell, called a CAR T cell, to the patient’s tumour when the cell is transferred back into the body. This therapy can be highly effective for tumours that express the protein CD19, such as B-cell acute leukaemias1,2 and large-cell lymphomas3,4. However, some people do not respond to CAR T cells, and efforts to optimize this therapy are ongoing. In a paper in Nature, Fraietta et al.5 report the fortuitous identification of a gene that positively affected one person’s response to treatment with CAR T cells.

Therapies involving engineered immune cells use viral vectors based on retroviruses or lentiviruses to insert a DNA sequence, such as one encoding a CAR, into a person’s T cells. However, given that there is no control over where the sequence inserts into the genome, it is possible that the engineered gene could insert at a location that disrupts another, important gene. In the early 2000s, a clinical trial6 enrolled people with immunodeficiencies arising from the lack of a functional copy of a particular immune gene. The trial used viral vectors to insert a wild-type copy of this gene into their stem cells. Unfortunately, however, several people developed uncontrolled T-cell proliferation that evolved into T-cell leukaemia. This event was linked7 to the gene inserting within the sequence of the LMO2 gene, disrupting the normal regulation of LMO2.

The pattern of genomic integration sites for various viral vectors has been found to be specific for a given combination of vector and cell type8. In a study of people who had T cells modified using retroviral vectors, the integration events were not implicated as the cause of any cancers9. Lentiviral vectors integrate randomly into the genome but tend to preferentially locate at sites of transcriptionally active genes10. Although random integration is generally thought to be safe, any disruption of the genome nevertheless confers a risk of unwanted consequences.

The effectiveness of treatments involving CAR T cells has been linked to the persistence and proliferation of the CAR T cells in the person’s body, and this can be affected by factors including the disease subtype, the molecular design of the CAR used, and even the manufacturing process1. Fraietta et al. report the unusual response of a person in a clinical trial whose CAR T cells targeted a CD19-expressing tumour called chronic lymphocytic leukaemia. In this case, disruption of the gene into which the CAR sequence had been inserted had a direct and beneficial effect on the clinical outcome.

The patient began to show a noticeable response to treatment two months after receiving a second dose of CAR T cells. Tumour regression normally occurs within a month if treatment is successful, so the authors investigated the reason for the delay in this case. Crucially, they analysed the nature of the CAR T cells at peak concentrations in the blood during tumour regression. Fraietta and colleagues made the surprising observation that these CAR T cells consisted almost exclusively of a clonal population descended from a single cell.

This single cell’s progeny divided over time until the cellular descendants reached a tipping point that eliminated the entire tumour. It is remarkable that the minimally effective and curative dose of this form of immunotherapy can be the introduction of just one cell. This raised the question of why introducing the CAR sequence to this specific T cell caused such an effective antitumour response.

In this clonal population of T cells, the CAR sequence had inserted into a copy of the TET2 gene, preventing the gene from encoding a functional protein. The patient’s other copy of TET2 had a mutation, so insertion of the CAR sequence generated T cells that lacked TET2 protein. TET2 is an enzyme, also called methylcytosine dioxygenase, that catalyses a hydroxylation reaction that alters methyl groups attached to DNA (Fig. 1). Such modifications of DNA or its associated proteins are known as epigenetic modifications, and they can affect gene expression in some cases. When Fraietta and colleagues compared the patient’s T cells that lacked the CAR insertion with those into which the CAR had been inserted, the overall epigenetic profile was similar. However, differences in the structure of the DNA–protein complex called chromatin were observed in genes involved in T-cell function, including CD28, ICOS and the gene that encodes interferon-γ.

Figure 1 | Tumour targeting by CAR T cells. If a patient’s T cells are engineered to express a version of an immune-cell receptor called a CAR, the cells can target tumour cells that express a specific protein, such as CD19. However, not everyone responds to this treatment. Fraietta et al.5 report that one patient’s response to CAR T-cell treatment has revealed a gene that can affect therapy success. The patient had a mutation in one of their copies of the TET2 gene. TET2 encodes an enzyme that converts methyl (CH3) groups attached to DNA into hydroxymethyl (CH2OH) groups. This type of change is known as an epigenetic modification. When a CAR-encoding sequence was introduced into the patient’s T cells using a viral vector, in one cell the CAR sequence inserted into the patient’s non-mutated copy of TET2 and disrupted the gene, thereby generating a cell that lacked any functional copies of TET2. The clonal descendants of this cell eradicated the patient’s tumour. The lack of TET2 altered the cell’s profile of epigenetic modifications, which can affect gene expression. This TET2 deficiency was associated with an increase in the expression of tumour-killing factors such as the enzyme granzyme, as well as entry into a cellular state called the central memory state, which stops the cells from entering a dysfunctional mode called exhaustion.

TET2 mutations have previously been associated with clonal blood-cell alterations linked to a risk of disease or blood cancers (a phenomenon known as clonal haematopoiesis)11. However, the patient’s T cells that lacked TET2 did not give rise to either aberrant T-cell proliferation or cancer. After tumour elimination, the number of CAR T cells decreased appropriately, replicating the normal pattern for a T-cell population (increasing in response to its target and declining after target elimination).

The authors used genetic engineering to remove TET2 in human T cells in vitro. Analysis of these cells revealed a connection between the absence of TET2 and the promotion and maintenance of T cells in a cellular state known as a central memory state. This state helps to prevent the cells from entering a dysfunctional mode called exhaustion, which is linked to ineffective tumour targeting by T cells. The absence of TET2 was also linked to an increase in long-term T-cell memory.

Fraietta et al. observed that human T cells lacking TET2 made fewer immune signalling molecules called cytokines than did cells that had TET2. Disruption of TET2 was also linked to an increase in the level of the enzymes perforin and granzyme, which are components of the tumour-killing machinery of T cells. These roles of TET2 in T-cell function and memory were previously unknown.

These remarkable findings might suggest that targeting TET2 in human T cells through drug-mediated inhibition or gene-editing techniques could increase the effectiveness of CAR T-cell treatment for other patients. If so, perhaps the dose of CAR T cells needed might be only a few cells, rather than the usual 50 million to 500 million cells. This would shorten the waiting time for CAR T cells and lower the substantial manufacturing costs. However, given the known associations of TET2 mutations with certain disease states, this approach might run the risk of generating a malignancy.

Enhancing CAR T-cell function is an area of active research, and other options to achieve this goal have been proposed. Inserting a CAR sequence at the genomic location where the natural version of the gene resides enhances the activity and persistence of CAR T cells in a mouse model12. Other groups have reported progress13 in making CAR T cells resistant to inhibitory checkpoint-signalling pathways that hinder T-cell function.

Will one of the many possible approaches be preferable to the others? Unfortunately, animal studies are not always predictive of results in humans, so clinical trials are the only way to answer this definitively. The good news is that it seems likely that many of these approaches will enhance efficacy and safety, so there is hope that the use of CAR T cells to treat cancer will become even more successful in the years to come.

Nature 558, 193-195 (2018)

doi: 10.1038/d41586-018-05251-5


  1. 1.

    Maus, M. V. & June, C. H. Clin. Cancer Res. 22, 1875–1884 (2016).

  2. 2.

    Park, J. H. et al. N. Engl. J. Med. 378, 449–459 (2018).

  3. 3.

    Neelapu, S. S. et al. N. Engl. J. Med. 377, 2531–2544 (2017).

  4. 4.

    Schuster, S. J. et al. N. Engl. J. Med. 377, 2545–2554 (2017).

  5. 5.

    Fraietta, J. A. et al. Nature 558, 307–312 (2018).

  6. 6.

    Hacein-Bey-Abina, S. et al. N. Engl. J. Med. 346, 1185–1193 (2002).

  7. 7.

    Hacein-Bey-Abina, S. et al. Science 302, 415–419 (2003).

  8. 8.

    Biasco, L. et al. EMBO Mol. Med. 3, 89–101 (2011).

  9. 9.

    Scholler, J. et al. Sci. Transl. Med. 4, 132ra53 (2012).

  10. 10.

    Schröder, A. R. W. et al. Cell 110, 521–529 (2002).

  11. 11.

    Buscarlet, M. et al. Blood 130, 753–762 (2017).

  12. 12.

    Eyquem, J. et al. Nature 543, 113–117 (2017).

  13. 13.

    Ren, J. et al. Clin. Cancer Res. 23, 2255–2266 (2016).

Download references

Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up