TCR-based Immunotherapy for Hematologic Malignancies

Unstructured abstract

Adoptive immunotherapy with engineered T cells is at the forefront of cancer treatment. T cells can be engineered to express T cell receptors (TCR) specific for tumor-associated antigens (TAAs) derived from intracellular or cell surface proteins. T cells engineered with TCRs (TCR-T) allow for targeting diverse types of TAAs, including proteins overexpressed in malignant cells, those with lineage-restricted expression, cancer-testis antigens, and neoantigens created from abnormal, malignancy-restricted proteins. Minor histocompatibility antigens can also serve as TAAs for TCR-T to treat relapsed hematologic malignancies after allogeneic hematopoietic cell transplantation. Moreover, TCR constructs can be modified to improve safety and enhance function and persistence of TCR-T. TCR-T therapies targeting three different TAAs are in early phase clinical trials for treatment of hematologic malignancies. Preclinical studies of TCR-T specific for many other TAAs are underway and offer great promise as safe and effective therapies for a wide range of cancers.

Introduction

The importance of T cells in cancer immunity is well-established. T cells recognize antigens that are short peptides derived from intracellular or cell surface proteins, presented in complex with major histocompatibility complex (MHC) molecules, also referred to as human leukocyte antigens (HLA) on human cells. Spontaneous T cell responses to a variety of cancer antigens have been observed, including in early murine models where exposure to viable or lethally irradiated tumor induced protective immunity against subsequent tumor exposure.¹ Endogenous T cell responses also occur in patients, but are often incompletely effective against advanced malignancies. Multiple mechanisms mediate tumor escape and/or block the formation of efficacious anti-tumor immune responses. These mechanisms include induction of a metabolically hostile tumor microenvironment, recruitment of suppressor cells, such as macrophages, regulatory T cells and myeloid-derived suppressor cells, production of immunosuppressive cytokines, expression of T cell inhibitory ligands by tumor or associated cells, and deletion of antigen-specific T cells.^2–4 Acute myeloid leukemia (AML), for example, produces an immunologically hostile environment⁵ through multiple mechanisms such as overexpressing indoleamine 2,3 dioxygenase 1⁶ and secreting arginase II⁷ to metabolically suppress T cells, overexpressing the PD-L1 T cell-inhibitory molecule,⁸ and blocking transcription factor activities needed for T cell activation and proliferation.⁹ Immunosuppressive microenvironments have also been observed in other hematologic malignancies, including chronic myeloid leukemia (CML)¹⁰ and chronic lymphocytic leukemia (CLL).¹¹ Consequently, T cells that could mediate anti-tumor immune responses become quantitatively or qualitatively defective. Adoptive transfer of T cells with tumor specificity is one approach to overcoming such deficiencies in endogenous anti-tumor immunity.

Adoptive T cell therapy for cancer treatment

To take advantage of naturally occurring T cell responses, one can isolate tumor-infiltrating lymphocytes (TIL) from tumor, activate and expand the T cells ex vivo, and re-infuse the resulting TIL product,^12–15 producing clinical responses in some patients. The technology is limited primarily to solid tumors with surgically accessible lesions, although marrow-infiltrating lymphocytes (MIL) are being tested as a therapeutic for hematologic malignancies.^16,17 However, neither TIL nor MIL can be generated for all patients.^18,19 Moreover, the specificities of infused TIL/MIL are generally undefined, and a minority of the T cells are tumor-reactive.²⁰ In contrast, T cells can be modified to express cell surface receptors that confer specific recognition of a malignant cell target, allowing administration of a T cell product with a defined specificity and composition.^21,22 Chimeric antigen receptors (CARs) are artificial antigen-specific receptors consisting of an extracellular ligand-binding domain linked to a CD3ζ chain along with one or more costimulatory domains^23,24 that can be transferred into T cells to produce an engineered T cell (CAR-T) with defined specificity for a cell surface molecule. Alternatively, T cells can be engineered to express a transgenic T cell receptor (TCR-T) specific for a tumor-associated antigen (TAA) formed from the complex of a tumor peptide and an MHC molecule (pMHC). Natural TCRs, from which transgenic TCRs are derived, consist of heterodimers of alpha and beta chains expressed in complex with CD3 proteins (Figure 1), and associate with CD4 or CD8 co-receptors during pMHC engagement and often with costimulatory molecules, such as CD28.

Figure 1.

Types of adoptive T cell therapy

Both CAR-T and TCR-T are adoptive cellular therapies with a defined, pre-selected target; both are often employed with a defined cell composition and can be generated for most patients in which the appropriate target is expressed. Each approach has advantages and disadvantages (Table 1). The major advantage for TIL/MIL and TCR-T over CAR-T is the ability to target peptides derived from intracellular proteins or cell surface proteins. Not all malignancies will have a suitable target for CAR-T therapy; that is, a surface molecule with high expression on tumor but low or absent expression on normal tissue, unless the normal tissue is dispensable. In contrast, TCR-T can allow access to the intracellular malignant proteome. CAR-T, however, are MHC/HLA-independent and therefore can be used in patients of all HLA types. CARs are expressed at higher surface levels than TCRs, but have less efficient signaling kinetics that do not properly recapitulate physiological TCR signaling leading to lower sensitivity to antigen.^25,26 Harris and colleagues engineered two TCRs with high affinity for two distinct TAAs and compared the function of the TCRs either as conventional αβ TCRs or as the ligand-binding domain of CAR constructs. Although the CAR constructs had similar binding affinity to αβ TCRs, primary T cells expressing the TCRs secreted cytokine in response to 100-fold lower peptide concentrations than T cells expressing the equivalent CAR.²⁵ Additionally, at high antigen density, CAR-T mediate greater maximal release of some cytokines, such as interleukin (IL)-2 and IL-6,²⁵ which can initiate a cascade leading to toxic cytokine release syndrome (CRS). TCR-Ts harness an endogenous T cell response that has been fine-tuned by the evolution of the immune system and are rarely associated with clinically significant CRS.

Table 1.

Approaches to adoptive T cell immunotherapies

	Advantages	Disadvantages
TIL	Polyclonal (reduce potential for escape through antigen loss); targets from intracellular and cell surface proteins	T cell specificity generally unknown; cannot be generated for all patients; not well established for hematologic malignancies; patient specific
TCR-T	Targets derived from intracellular and cell surface proteins; defined specificity	HLA restriction; tumor escape through antigen loss (altered processing of peptides, HLA downregulation)
CAR-T	No HLA-restriction; defined specificity	Cell surface targets only; tumor escape through antigen loss (downregulation or loss of target protein)

Antigen targets of T cell receptors for immunotherapy

Antigen processing and presentation

T cell antigens are generated from full-length proteins through a multi-step intracellular process (Figure 2) that has been extensively reviewed.^27–29 CD8⁺ T cell antigens are primarily derived from endogenous proteins that are degraded in the cytoplasm by proteasomes and cytosolic aminopeptidases. The resulting peptides are transported into the endoplasmic reticulum (ER) through the transporter associated with antigen processing (TAP) complex, where they are trimmed by ER aminopeptidases (ERAPs) into lengths of 8–10 amino acids. While we have a detailed understanding of how processing occurs, the exact rules of trimming and processing are not well understood. Trimmed peptides are assembled with MHC class I heavy chain and beta-2 microglobulin via a peptide-loading complex. The pMHC class I complex then transits from the ER to the plasma membrane, where it is presented externally. CD8⁺ T cell antigens can also be generated from viral proteins in infected cells and from misfolded or improperly synthesized proteins. Additionally, certain antigen-presenting cells can present peptides from internalized exogenous proteins on MHC class I molecules through cross-presentation. TAA are potentially present on all malignant cells, regardless of the tissue of origin, because MHC class I molecules are ubiquitously expressed on all nucleated cells. However, downregulation and loss of class I MHC expression are established mechanisms of immune evasion by tumors.^30–38 CD8⁺ T cells are directly cytotoxic, making MHC class I restricted TAA particularly attractive as immunotherapy targets.

Figure 2.

Antigen processing and presentation

In contrast, CD4⁺ T cells recognize peptides presented on MHC class II molecules. Peptides are loaded onto MHC Class II after degradation of internalized exogenous proteins or autophagy of endogenous proteins.³⁹ Class II molecules are normally expressed on antigen-presenting and hematopoietic cells, including AML. This review will focus on class I-restricted CD8⁺ T cell antigens, since these are the majority of targets in preclinical and clinical development for TCR-T therapies for hematologic malignancies. However, class II restricted antigens also appear to have a role in antitumor immunity. Downregulation of MHC class II expression⁴⁰ or complete loss of a mismatched MHC haplotype⁴¹ can occur in leukemic relapses after allogeneic hematopoietic stem cell transplantation (allo-HCT), suggesting that class II antigens contribute to disease control after allo-HCT. Upregulation of class II expression been described for some solid tumors,^42–45 and, in melanoma, is associated with increased probability of response to immune checkpoint blockade therapy.⁴⁶ Adoptive transfer of TIL enriched for CD4⁺ T cells specific for a mutation in erbb2-interacting protein produced striking clinical responses in one patient with metastatic cholangiocarcinoma.⁴⁷ As we learn more about class II-restricted TAAs, these too may prove to be relevant targets in hematologic malignancies.

Classes of T cell antigens for TCR-based immunotherapy

An ideal target for any cell-based immunotherapy is one that is selectively presented on malignant cells, necessary for survival of the tumor cells, and shared amongst patients. Four major TAA classes have been considered as targets for TCR-T immunotherapy in solid and liquid tumors: 1) overexpressed antigens derived from wild-type proteins with relatively high expression in malignant cells; 2) lineage-restricted antigens that are also presented on the normal counterparts of the malignant cells; 3) cancer-testis antigens (CTA), which are normally expressed in germline tissues and aberrantly expressed in malignant cells; and 4) neoantigens created from abnormal proteins (mutations, fusions, frameshifts, or novel isoforms) or abnormal peptides specific to the malignant cells. Antigen specificity for malignant versus healthy cells varies among the categories of TAAs (Table 2), and the potential risk of on-target, off-tumor toxicity inversely corresponds to that specificity. Specific examples of TAAs in each category are shown in Table 3.

Table 2.

Categories of TAAs for the development of TCR-T immunotherapy

	Overexpressed	Lineage-restricted	Cancer-testis antigens	Neoantigens	Minor histocompatibility (H) antigens
Antigen derived from	Wild-type protein with relatively increased expression in malignant cells	Wild-type protein expressed in malignant cells and normal counterparts	Wild-type protein expressed only in malignant cells and germline cells	Abnormal protein created by cancer-specific mutation, gene fusion, frame shift or abnormal splicing, or peptide created by abnormal antigen processing	Polymorphic normal peptides created by polymorphisms that differ between donor and recipient in allogeneic hematopoietic cell transplantation (HCT)
Specificity for malignant cells	Lowest	Low	Moderate	Highest	Hematopoietic-resetricted minor H antigens become leukemia-specific following allo-HCT
Potential for on-target off tumor toxicity	High	Moderate/high (may be acceptable if normal counterpart is dispensable)	Low (Testis/germline lissues lack HLA)	Lowest	Variable
Breadth of potential applicability	High	High	Moderate/high	Low	Moderate/high (only for relapses after allo-HCT)

Table 3.

Examples of TAA in hematologic malignancies and solid tumors

Category of TAA	Cancer types	Examples	Diseases	Other tissues	References
Overexpressed	Hematologic	WT1	AML, MDS	Kidney (podocytes), CD34+ cells	91
		Survivin	AML, ALL, MDS	None in adult tissues	3
		hTERT	ALL, AML, CLL	Ovaries, testis	113–115
		BOB1	B cell leukemias and lymphomas, multiple myeloma	Normal B cells	121
Lineage-restricted	Hematologic	CD20	Lymphoma	Normal B cells	122
		CD22	ALL		123
	Solid tumor	MART-1/MelanA	Melanoma	Normal melanocytes	141,142
		Tyrosinase			Reviewed in 143
		gp100
Cancer-testis antigens	Hematologic	PRAME	AML, ALL, MDS, multiple solid tumors	Adrenals, ovaries, endometrium, testis	144,102
		Aurora kinase A	AML, CML, ATL	Testis	125,126 127
	Solid tumor	MAGE family	Melanoma, lung, breast, colorectal, prostate	Testis, brain (MAGEA12)	Reviewed in 143
Neoantigens	Hematologic	NPM1	AML	None	130
		CBFB-MYH11			129
Leukemia-associated minor histocompatibility (H) antigens	Hematologic	HA-1	Relapsed hematologic malignancies after allo-HCT	Hematopoietic	69

In hematologic malignancies, minor histocompatibility (H) antigens are a fifth important class of TAAs. Minor H antigens are MHC/HLA-bound polymorphic peptides that differ between allo-HCT recipient and donor as a result of genetic polymorphisms. Once full donor normal hematopoietic chimerism is achieved after allo-HCT, hematopoietic-restricted minor H antigens are present only on residual recipient malignant hematopoietic cells, providing significant specificity when donor and recipient are mismatched for the polymorphism. Like lineage-restricted antigens, the specificity of minor H antigens for malignant cells depends on how tightly restricted expression of the parent protein is to hematopoietic cells. Therapies targeting minor H antigens could potentially treat multiple hematologic malignancies, since the antigens are not disease specific. However, varying polymorphism frequencies across populations and the requirement for donor-recipient mismatch currently limits the applicability of minor H-directed immunotherapy to a subset of individuals who relapse after allo-HCT.

General considerations in developing TCR immunotherapy

Transgenic TCR development begins by discovering and cloning a naturally occurring TCR specific for a suitable target. Generally, there are three starting pools of cells in which to identify a relevant TCR: patient TIL/MILs, patient peripheral blood T cells, and healthy donor peripheral blood T cells requiring primary in vitro stimulation.^48,49 TIL/MILs can be used as a source of T cells potentially enriched for TAA-specific TCRs, but T cells derived from immunosuppressive tumor environments are often dysfunctional.⁵⁰ Stimulating healthy donor T cells with known or predicted novel TAAs in vitro to isolate reactive T cells and their TCRs can circumvent T cell dysfunction.

To determine reactivity to a specific target, binding and functionality are tested by measuring pMHC multimer binding, cytotoxicity, and/or cytokine production by ELISPOT or flow cytometry. Newer methods like barcoded dextramer staining⁵¹ allow screening hundreds of pMHC complexes from one sample. Once responding T cells are found, the TCR sequence must be determined and cloned. Rapid amplification of cDNA ends before polymerase chain reaction (RACE-PCR) allows identification of the TCR α and β chains of a reactive T cell clone. Contemporary approaches like single-cell sequencing can directly identify the TCR sequences of individual clones in a bulk population. The Wu group paired single-cell sequencing with a library of cloning plasmids for each TCRα/TCRβ chain to reconstruct TCRs from a bulk population and rapidly deconvolute TCR specificity and avidity for a target antigen,⁵² allowing TCRs to be screened against relevant targets at an accelerated pace compared to traditional methods.

After sequencing a reactive TCR, the receptor must be transferred to a new T cell. Transduction with a viral vector encoding a polycistronic construct of both α and β chains of the transgenic TCR separated by a ribosomal skip motif, such as a 2A self-cleaving peptide, is most commonly used. Nonviral techniques for TCR gene transfer, such as transposon-based technologies^53–55 and nanoparticles,⁵⁶ are also in development. In any case, conventional T cells that are genetically modified with a transgenic TCR have their own endogenous TCRs, making mispairing of the transferred and endogenous TCRs a concern. In preclinical studies using an OT-I TCR murine model, mispairing of the endogenous TCR with the transferred TCR led to lethal graft-versus-host disease (GVHD),⁵⁷ although GVHD due to TCR mispairing has not been observed in human trials to date. The risk of mispairing can be mitigated by introducing cysteine modifications to the transferred TCR α and β chains to favor pairing of the introduced chains to each other. Other potential modifications to the TCR construct include codon optimization and minimal murinization of the constant region⁵⁸ to enhance expression of the transgenic TCR and encourage correct pairing of the introduced chains. Alternatively, using small interfering RNA (siRNA) and CRISPR/Cas9 to disrupt the endogenous TCR has been shown to limit toxicity and increase T cell activity in humanized mouse models.⁵⁹ While eliminating the endogenous TCR prevents mispairing, it is unclear whether retaining an endogenous TCR might help transferred T cells persist when antigen burden is low. However, it is clear that designing transgenic TCRs for exclusive self-pairing is paramount to limiting autoimmunity and toxicity.

Many native TCRs recognize overexpressed self-antigens with inherently low affinity due to tolerance mechanisms. In vitro affinity maturation can enhance target recognition,^60–62 although with increased risk of cross-reactivity and off-target toxicity. In a clinical trial of an affinity-matured MAGE-A3-specific TCR, transgenic T cells also recognized the Titin cardiac protein and led to the death of two patients.⁶³ The native, non-matured TCR did not recognize cardiac tissue,⁶⁴ highlighting the relationship between augmented affinity and potential cross-reactivity. Which complementary-determining regions (CDRs) are altered in affinity maturation may influence the risk of cross-reactivity and toxicity. CDR1 and CDR2 interact predominantly with MHC; alterations in these regions can increase TCR binding of MHC regardless of peptide.^65,66 In contrast, CDR3 interacts predominantly with peptide,⁶⁷ so affinity maturation targeting this region might produce higher affinity TCRs with less likelihood of cross-reactivity, although experimental data is limited. A recent approach to developing TCRs with highly diverse CDR3s used in vitro antigen-driven differentiation of progenitor T cells to generate higher affinity TCRs without changes to CDR1/2.⁶⁸ Safety testing of TCRs generated in this manner may lead to better understanding of the relationship between cross-reactivity and affinity.

As our understanding of natural and synthetic T cell biology increases, investigators are testing numerous modifications that might improve the functionality, persistence, and safety of transferred T cells (Figure 4). Introducing a CD8 co-receptor can facilitate recognition of class I antigens by CD4⁺ T cells, and enhance ‘help’ for cytotoxic CD8⁺ T cells.^69–74 Safety switches, including inducible pro-death proteins, can be used to rapidly remove transferred T cells if toxicities such as CRS or GVHD become an issue.^75–80 The ability to secrete IL-12 and other cytokines can increase activity of so-called ‘armored’ CAR-T or TCR-T cells and create a proinflammatory environment that enables antigen presentation,^81–83 albeit with an increased potential for toxicity if cytokine secretion is not tightly regulated.⁸⁴ Function of transferred T cells can also be increased by siRNA or CRISPR-Cas9 elimination of inhibitory molecule genes, such as PD-1,^85,86 or by converting negative malignant cell-derived signals into activation signals for engineered T cells, for example by fusing the inhibitory receptor CD200R to a costimulatory CD28 domain.⁸⁷ Normal signaling pathways, such as thrombopoietin and c-MPL,⁸⁸ can also be co-opted to deliver an activating signal to engineered T cells through the transgene construct. ‘Off-the-shelf’ versions of engineered cells that are HLA negative and express natural killer (NK) cell inhibitory molecules could be used as universal donor cells,⁸⁹ potentially eliminating the need for autologous cell collection from patients, although silencing the endogenous TCR will likely be required in this situation to prevent GVHD.

Figure 4.

Construct modifications to enhance TCR-T cells

TCR immunotherapy in hematologic malignancies

A recent systematic review estimated that only 16% of previous and current TCR-T clinical trials were aimed at treating hematologic malignancies.⁹⁰ However, numerous preclinical studies are presently underway. Suitable TCR targets must be highly expressed on malignant cells, including blasts, progenitor and cancer stem cells, and have little or ideally no expression on healthy non-hematopoietic tissues. However, expression of the target antigen on normal hematopoietic cells may be acceptable in select circumstances; for example, if: 1) expression is sufficiently low that high-avidity T cells will not recognize normal cells; 2) the normal cells are relatively dispensable, as in the case of normal B cells; 3) the normal cells do not express HLA molecules (e.g. testis); or, 4) TCR immunotherapy expected to cause myelolablation will be used prior to allo-HCT and the TCR-T cells are designed to be short-lived. Because the field is still evolving and the optimal antigens for TCR immunotherapy are as-yet unknown, target candidates from all TAA categories are being explored (Table 3).

Clinical development of TCR immunotherapies

WT1 TCR immunotherapy

Wilm’s tumor 1 (WT1) protein is overexpressed in acute leukemias and myelodysplastic syndromes (MDS)⁹¹ and has limited expression on normal tissues, including normal CD34⁺ hematopoietic stem cells. T cells specific for WT1 epitopes that are presented on HLA-A*24:02⁹² or -A*02:01^93,94 recognize primary leukemic blasts, making WT1 an attractive therapeutic target. Donor-derived T cell responses to WT1 have been observed in vivo after allo-HCT.⁹⁴ WT1-specific CD8⁺ T cell responses can be stimulated ex vivo, and retain their activity in vivo.^95,96 The Chapuis and Greenberg group adoptively transferred ex vivo expanded, donor-derived CD8⁺ T cells specific for an HLA-A*02:01-restricted WT1 epitope in a cohort of eleven patients with relapsed or high-risk acute leukemia or MDS after allo-HCT (NCT00052520).⁹⁷ Transferred T cells were well tolerated, indicating minimal on-target off-tumor toxicity, persisted and showed direct (disease response) or indirect (absence of relapse) anti-leukemic activity in five patients. Other clinical trials of ex vivo expanded or sensitized WT1-specific T cells for high-risk leukemias and multiple myeloma are ongoing (NCT00620633, NCT01758328).

Chapuis, Greenberg and coworkers next developed transgenic TCR-T cells directed against an HLA-A*02:01-restricted WT1 epitope for prevention or treatment of AML relapse after allo-HCT. To generate WT1-specific T cells, cytomegalovirus- or Epstein Barr virus-specific T cells from the HCT donor were transduced with a native, high-affinity, WT1-specific TCR identified from the peripheral repertoires of a healthy HLA-A*02:01⁺ individual. In a phase 1 clinical trial (NCT01640301), eleven patients with high-risk AML were treated prophylactically with WT1 TCR-T cells, and none had relapsed at a median follow-up of 21.3 months after allo-HCT, compared to 27% relapse at 16 months among matched controls.^98,99 No survival advantage over standard of care was seen in patients treated with WT1 TCR T cells at relapse in preliminary findings. The Tawara group developed a retroviral construct encoding a high-affinity WT1/HLA-A*24:02-specific TCR identified from the peripheral repertoire of a healthy individual,⁹² along with siRNAs to eliminate expression of endogenous TCR chains.¹⁰⁰ In a phase 1 clinical trial (UMIN000011519), eight patients with chemotherapy-refractory AML or high-risk MDS were treated with escalating doses (1.2–3.5× 10⁸ cells per infusion) of autologous T cells expressing the WT1 TCR construct. No toxicity was observed, WT1 TCR T cells persisted in five patients, and transient decreases were noted in the percentage of bone marrow leukemic blasts in two patients. These data suggest that adoptively transferred WT1 TCR T cells directed against either HLA-A*02:01 or -A*24:02-restricted WT1 epitopes are safe, well-tolerated and may have anti-leukemic activity in vivo.

PRAME TCR immunotherapy

The PRAME CTA is overexpressed on solid tumors and in acute lymphoblastic leukemia (ALL), AML and MDS.^101,102 A high-avidity TCR specific for an HLA-A*02:01-restricted, PRAME-derived epitope was identified from the T cell repertoire of an HLA-A*02:01-negative donor after transplantation and subsequent donor lymphocyte infusion into an HLA-A*02:01-positive recipient.¹⁰³ The parental clone from which this TCR was isolated could recognize primary AML and ALL samples, as well as solid tumor cell lines, but did not recognize normal tissues except for mature dendritic cells and kidney epithelial cells. Retroviral transfer of the TCR conferred functional avidity and recognition of target cell lines similar to that measured for the parental clone. An early phase clinical trial of adoptive transfer of PRAME/A*02:01 TCR-transduced autologous T cells (BPX-701, Bellicum Pharmaceuticals) to treat AML, MDS and uveal melanoma opened in 2017 (NCT02743611). A second PRAME/A*02:01-specific TCR was isolated independently by another group;¹⁰⁴ an early phase clinical trial testing autologous T cells expressing this TCR for treatment of high-risk AML, MDS and multiple myeloma opened in 2017 in Germany (NCT03503968).

HA-1 T cell immunotherapy

Hematopoietic-restricted minor H antigens can function as TAA in the context of recurrence after allo-HCT, assuming that donor and recipient have a suitable genotype mismatch. The minor H antigen HA-1^H (hereafter referred to as HA-1) is encoded by a DNA sequence spanning a single nucleotide polymorphism (RS_1801284) within the HMHA1 gene. The resulting immunogenic peptide, VLHDDLLEA, is efficiently presented by HLA-A*02:01; the corresponding allelic variant peptide, VLRDDLLEA, has lower affinity for and unstable binding to HLA-A*02:01. The HMHA1 protein product is expressed in normal and malignant hematopoietic cells, but not in non-hematopoietic cells. In HA-1-mismatched allo-HCT, high-avidity HA-1-specific T cells from a HA-1-negative donor can be primed against HA-1 and mediate a graft-versus-tumor effect, as evidenced by decreased relapse rates in HA-1 mismatched allo-HCT (donor negative, recipient positive) and expansion of HA-1-specific T cells after donor lymphocyte infusion for post-allo-HCT relapses.¹⁰⁵ A high-avidity, HA-1-specific TCR was identified from the repertoire of a healthy HLA-A*02:01-positive HA-1-negative individual and was sequenced and cloned into a lentiviral vector, along with a selection marker, safety switch (iCasp9) and CD8 co-receptor, to allow TCR-transduced CD4⁺ T cells to recognize the MHC class I-restricted epitope. CD8⁺ and CD4⁺ T cells expressing the HA-1 TCR construct were highly functional against leukemia cell lines and primary leukemia in vitro and could be rapidly eliminated using the safety switch.⁶⁹ A phase 1 clinical trial of CD8⁺ and CD4⁺ T cells transduced with this HA-1 TCR construct is now open to patients with hematologic malignancies (NCT03326921). HA-1 TCR T cell immunotherapy is also being evaluated in the Netherlands, using a different HA-1-specific TCR, a different transgene (without CD8 coreceptor) and viral vector (retroviral rather than lentiviral), and a different cell product (EudraCT number 2010–024625–20).

Preclinical development of TCR immunotherapies

Overexpressed antigens

Survivin

Survivin is a member of the inhibitors of apoptosis protein family that is overexpressed in numerous cancers, including AML^106–108 and various lymphomas,^109,110 where its overexpression correlates with poor prognosis. High-affinity TCRs specific for an HLA-A*02:01-restricted survivin epitope were identified by stimulating HLA-A*02:01-negative CD8⁺ T cells with autologous dendritic cells transfected to express HLA-A*02:01. While transfer of TCRs isolated by this method enabled recognition and lysis of survivin-positive tumor cell lines, transfer of high-affinity TCRs also produced HLA-restricted fratricide of HLA-A*02:01-positive T cells.¹¹¹ To overcome this obvious limitation, another group subsequently used stimulation with autologous dendritic cells to identify survivin-specific, HLA-A*02:01-restricted T cells from HLA-A*02:01-positive individuals.¹¹² Adoptive transfer of T cells transduced to express survivin-specific TCRs thus isolated showed antileukemic activity and prolonged survival in immunodeficient murine xenograft models of AML.¹¹²

Human telomerase reverse transcriptase (hTERT)

hTERT is a catalytic subunit for telomere elongation that is expressed in numerous hematologic malignancies, but not in normal tissues.^113–115 Two hTERT epitopes have been investigated as targets for transgenic TCR immunotherapy. A high-avidity TCR specific for an HLA-A*24:02-specific hTERT epitope¹¹⁶ was identified from the repertoire of a healthy individual by autologous stimulation, and subsequently cloned into a retroviral vector that also included siRNAs to eliminate expression of endogenous TCR chains.¹¹⁷ Adoptive transfer of hTERT TCR-T cells into an immunodeficient murine xenograft model of HTLV-associated T cell leukemia (ATL) inhibited tumor outgrowth for >6 months. In separate studies, a murine TCR specific for an HLA-A*02:01-restricted hTERT epitope was isolated by hTERT vaccination of HLA-A*02:01-expressing transgenic mice¹¹⁸ and cloned into a retroviral vector.¹¹⁹ Adoptive transfer of human T cells transduced with the hTERT-specific TCR showed in vivo activity against CLL¹¹⁹, B cell ALL, and AML¹²⁰ in murine models. Although we are not aware of any TCR-T clinical trials targeting hTERT, clinical trials of hTERT vaccination are underway in the U.S. and Europe.

BOB1

BOB1 is a B cell-specific transcription factor that is highly expressed in B cell leukemias and lymphomas, as well as in multiple myeloma. A TCR specific for an HLA-B*07:02-restricted BOB1 epitope was identified from the alloreactive repertoire of a healthy individual and transferred into a retroviral vector.¹²¹ The transferred TCR enabled selective recognition and killing of BOB1-positive primary B cell leukemia, mantle cell lymphoma and multiple myeloma cells in vitro. Adoptive transfer of BOB1 TCR-transduced CD8⁺ T cells also controlled tumor outgrowth in an immunodeficient murine xenograft model of established myeloma.¹²¹

Lineage-restricted

Targeting lineage-restricted antigens in myeloid malignancies can cause myeloablation, necessitating allo-HCT to rescue hematopoiesis. However, in B cell malignancies, targeting of lineage-restricted antigens is more feasible because healthy B cells are dispensable. While this capacity has been exploited most thoroughly in the development of CAR-T cell therapy (CD19-, CD22-, and CD20-directed), B cell-restricted antigens can also be targeted by TCR-based approaches. Transfer of an HLA-B*07:02-restricted CD20 epitope-specific TCR enabled recognition of CD20-positive malignant cell lines and primary tumors (CLL, ALL, and mantle cell lymphoma), and suggested that TCR-based approaches may be effective in low expression CD20-positive B cell malignancies.¹²² T cells transduced with a different TCR specific for an HLA-B*07:02-restricted CD22 epitope killed primary B cell leukemia cells, but also killed healthy dendritic cells and macrophages,¹²³ illustrating both the potential promise and limitation of targeting lineage-restricted antigens.

Cancer-testis antigens

Aurora kinase A (AurA) is a serine-threonine kinase involved in mitotic cell division, which among normal tissues is only detected in testis only, but is expressed at high levels in leukemias.^124,125 An epitope derived from amino acids 207–215 of AurA is presented on HLA-A*02:01 and -A*24:02 on primary myeloid leukemia, and CD8⁺ T cells specific for this epitope lyse leukemia cells in vitro.¹²⁵ A TCR specific for the HLA-A*02:01-restricted AurA epitope was identified in a healthy donor, and retroviral transfer of this TCR enabled in vitro recognition of leukemia by transgenic CD4⁺ and CD8⁺ T cells.¹²⁶ The CD8⁺ AurA TCR-T cells also controlled leukemia progression in vivo in an immunodeficient murine xenograft model. Of note, AurA TCR-T cells may also be applicable as immunotherapy in ATL.¹²⁷

Neoantigens

In contrast to solid tumors, which may carry hundreds or even thousands of mutations in an individual patient,¹²⁸ most hematologic malignancies have relatively few protein-coding mutations or gene fusions, and thus fewer potential neoantigens. However, for TCR-T immunotherapy, a small number of neoantigens will suffice, as long as those neoantigens are shared among patients and ideally occur early in oncogenesis or are essential in maintaining the malignant phenotype. For example, the type A variant of the CBFB-MYH11 fusion is critical in leukemogenesis, and occurs in ~10% of individuals with AML and ~90% of AML patients with the inv(16) or t(16;16) cytogenetic abnormalities. An epitope spanning the CBFB-MYH11 fusion region is presented on HLA-B*40:01 and enables selective recognition and killing of leukemic blasts by epitope-specific T cells. Lentiviral transfer of a high-avidity TCR confers epitope specificity and antileukemic cytotoxicity in vitro, suggesting that CBFB-MYH11 is a viable target for TCR-based immunotherapy for a subset of patients.¹²⁹ Targeting the fusion may enable eradication of the cell of origin, as it is known to play a key role in maintaining the leukemic phenotype. An epitope derived from a highly recurrent frameshift mutation in exon 12 of nucleophosmin1 (NPM1) is another potential neoantigen target for TCR immunotherapy. NPM1 mutations occur in 30–35% of AML cases and are early events that persist across the disease course. CD8⁺ T cells specific for an HLA-A*02:01-restricted NPM1 epitope can selectively recognize NPM1-mutated leukemic blasts, and adoptive transfer of NPM1/HLA-A*02:01 specific TCR-transduced T cells controlled tumor outgrowth and prolonged survival in an immunodeficient murine xenograft model.¹³⁰ These early studies suggest a role for shared neoantigens in TCR-based immunotherapy of AML and other hematologic malignancies.

Future of TCR immunotherapy in hematologic malignancies

TCR-T cell immunotherapy is a very young field but is evolving quickly. Advances in the fields of basic immunology, protein science, synthetic biology, genomics, and cell and genome engineering should allow us to overcome many previously-recognized obstacles and facilitate the development of TCR-T therapies. One major bottleneck has been identifying bona fide target epitopes with sufficient cancer-specificity for safe targeting in patients. In silico algorithms can effectively predict binding of peptides to MHC/HLA molecules, but not whether the peptides are processed and presented on cell surfaces. With the wave of ‘omic’ approaches, some have attempted to determine the ‘peptidome’ of malignant cells by immunoprecipitating MHC complexes, eluting peptides from these complexes and identifying peptides by tandem liquid chromatography–mass spectrometry.^131–133 This unbiased approach can identify peptides naturally processed and presented on cells of interest, but also has technical hurdles.^134–136 Improving techniques to identify and characterize the peptidome of malignant cells, such as the use of monoallelic cells,¹³⁴ has the potential to fast-track discovery of physiologically-relevant targets for TCR-T immunotherapy. TCR sequencing techniques are rapidly evolving, as is our understanding of how to predict which pMHC complexes functionally engage specific TCR sequences.¹³⁷ Although these technologies are early in development, understanding the biophysical and structural rules that govern TCR-pMHC binding will also improve prediction of cross-reactivity to other pMHC complexes and thus uncover potential off-target toxicities, and may even allow rational design of synthetic target-specific TCRs in the future.

Currently, all TCR-T immunotherapies target single antigens, running the risk the cancer may simply escape recognition by downregulating expression of the target protein, as seen in CD19 CAR-T cell trials.^138,139 TCR-T targeting of proteins that are essential to maintaining the leukemic phenotype, like leukemia-initiating fusions, should avoid this escape mechanism. For targets that are less essential to the malignant cell, transferring transgenic TCR-T cells with multiple specificities could reduce the probability of antigen escape. One approach would be to make multiple TCR-T products with different specificities separately and infuse them together into the patient. Another would be to modify T cells to express multiple TCRs, although ensuring correct pairing of introduced TCR chains would be challenging. TCR mimic monoclonal antibodies, which bind specifically to pMHC complexes, can be used as CARs¹⁴⁰ and might be more readily multiplexed.

Conclusions

Much progress has been made in the field of TCR-T immunotherapy generally and for hematologic malignancies specifically. Despite the challenges of identifying suitable targets for TCR-T therapies, a number of promising TAAs are under preclinical investigation as targets for TCR-T immunotherapy. Encouragingly, four different TCR-T immunotherapies are now in early phase clinical trials, targeting: WT1 epitopes restricted to HLA-A*02:01 and A*24:02; an HLA-A*02:01-restricted PRAME epitope; and the HLA-A*02:01-restricted minor H antigen HA-1. Along with improvements in T cell antigen discovery and cell engineering techniques, gained in part from experience with CAR-T cell therapies, these studies set the stage for development of TCR-T immunotherapies with potent curative potential in the hematologic malignancies.

Figure 3.

Categories of tumor-associated antigens