In this issue of Blood, Qin et al demonstrate the ability of chimeric antigen receptor (CAR)-engineered T cells targeted against thymic stromal lymphopoietin receptor (TSLPR) to eradicate disease in several models of B-cell acute lymphoblastic leukemia (B-ALL) that overexpress this protein.1
TSLPR-specific CAR. (A) A schematic diagram of a CAR expressed at the T-cell surface, illustrating the typical component parts found in a second-generation variant. (B) Cartoon depicting the elimination of B-ALL target cells with a CRLF2 rearrangement by TSLPR-retargeted CAR T cells. By contrast, healthy B-lymphocyte precursors (hematogones) express TSLPR at low levels, which may protect them from immune attack. Some epithelial cell types also express low levels of TSLPR, although immunohistochemical analysis suggests that the target is not present at the cell surface.
TSLPR-specific CAR. (A) A schematic diagram of a CAR expressed at the T-cell surface, illustrating the typical component parts found in a second-generation variant. (B) Cartoon depicting the elimination of B-ALL target cells with a CRLF2 rearrangement by TSLPR-retargeted CAR T cells. By contrast, healthy B-lymphocyte precursors (hematogones) express TSLPR at low levels, which may protect them from immune attack. Some epithelial cell types also express low levels of TSLPR, although immunohistochemical analysis suggests that the target is not present at the cell surface.
Chimeric antigen receptors are genetically delivered fusion proteins that redirect the specificity of polyclonal T cells or natural killer cells against a chosen cell surface molecule. Constructs consist of a targeting moiety such as a peptide, ligand derivative, or antibody-derived single-chain variable fragment (scFv) that is coupled in series to a hinge/spacer, membrane-spanning element, and signaling endodomain (see figure, part A). Target antigen is engaged in its native conformation, rather than as a processed peptide displayed within the groove of a human leukocyte antigen (HLA) molecule. Consequently, CARs recognize target cells irrespective of a patient’s HLA haplotype. Furthermore, their function is unhindered by a common immune evasion strategy that is deployed in acute lymphoblastic leukemia (ALL) and other malignancies, namely, the selective downregulation of some HLA allo-specificities. In so-called “first-generation” CARs, the endodomain generally contains CD3ζ alone, thereby providing signals that mimic those naturally provided by the T-cell receptor/CD3 complex. Second- and third-generation receptors are distinct in that they respectively contain either 1 or 2 additional costimulatory motifs, commonly derived from CD28 and/or tumor necrosis factor receptor families. Upon target ligation, integrated signaling by the CAR endodomain results in T-cell activation and target-cell killing (see figure, part B). The provision of costimulation in later-generation CARs ensures improved T-cell persistence in vivo,2 a property that correlates strongly with enhanced efficacy.
Target selection is of fundamental importance in maximizing the therapeutic index of CAR-based immunotherapy. Few tumor-associated antigens are truly disease specific, mandating careful risk assessment and vigilance for the emergence of “on-target, off-organ” toxicity. In B-ALL, unprecedented complete remission rates approaching 90% have been repeatedly attained in phase 1 clinical trials involving CAR T cells that engage the ubiquitous B-cell antigen, CD19.3 Because CD19 is expressed throughout B-cell differentiation, this success occurs at the expense of predictable toxicity in the form of B-cell aplasia and hypogammaglobulinemia. Although this is clearly undesirable, it can be managed effectively with IV or subcutaneously administered immunoglobulin replacement therapy. Of greater concern, however, is the risk of therapeutic failure or relapse with CD19 null disease.4 The latter may reflect the lack of direct contribution of CD19 expression to disease pathogenesis in B-ALL and highlights the need to identify additional candidate target molecules.
Expression of the cytokine receptor–like factor 2 (CRLF2) gene may be perturbed by a number of genomic alterations in B-ALL. As a result, enhanced cell-surface expression of the encoded TSLPR protein is seen in as much as 15% of cases that lack typical chromosomal rearrangements.5,6 This observation raises the possibility that TSLPR could represent an attractive new candidate for CAR T-cell therapy of some patients with B-ALL. In support of this, Qin et al demonstrate that T cells expressing a CAR targeted with a 3G11-derived scFv can be efficiently redirected against TSLPR+ target cells. Potent antitumor effects were observed both in vitro and in vivo, even in the setting of established disease. Complete tumor eradication was also demonstrated in several patient-derived xenograft models, confirming efficacy in a more clinically relevant setting. Importantly, T cells expressing the TSLPR-specific CAR demonstrated in vivo efficacy comparable with CD19-specific CAR T cells, thereby further confirming the potential of TSLPR as a target in B-ALL associated with CRLF2 rearrangements. In agreement with clinical findings, efficacy was associated with enhanced T-cell survival post–adoptive transfer. This was demonstrated by the fact that T cells expressing a CAR variant with a longer spacer exhibited satisfactory function in vitro, but failed to persist in vivo or to mediate antileukemic activity.
Although upregulation of TSLPR is restricted to a subset of patients, the demonstration that it can be targeted effectively using CAR-engineered T cells is important for 2 additional reasons. First, patients with B-ALL bearing CRLF2 rearrangements and TSLPR overexpression have particularly poor relapse-free and overall survival rates.7 Second, there is significant evidence that TSLPR contributes directly to disease pathogenesis in B-ALL, a factor that may protect against antigen loss. Overexpression of TSLPR in lymphoid progenitors is sufficient to promote their growth.5 Moreover, patient samples that express high levels of this receptor proliferate in response to TSLP, unlike samples lacking CRLF2 rearrangements.8 However, it should be noted that leukemias in which TSLPR is upregulated also exhibit a number of cooperative genetic changes, most notably activating mutations in Janus kinase 2 (JAK2).9 Although expression of mutated JAK2 was insufficient to enable immortalized Ba/F3 cells to grow in the absence of exogenous cytokine,6 it has been suggested that the signaling through mutated JAK2 may be responsible for the continued, albeit slower, growth of a B-ALL cell line in which TSLPR expression had been silenced.5 Consequently, it remains to be determined whether signaling provided by mutated JAK2 or related kinases will be sufficient to allow TSLPR loss in response to the selective pressure mediated by CAR T cells in vivo.
An additional caveat to the potential use of TSLPR as an immunotherapeutic target is the lower-level expression of this receptor in a number of other cell types, notably CD4+ T cells and dendritic cells.1 Although the level of TSLPR required for activation of the CAR+ T cells is unknown, the potential for on-target, off-organ toxicity is of concern and warrants close attention during phase 1 clinical evaluation. The CAR tested in this study did not crossreact with the mouse ortholog of TSLPR. This contrasts with some other CARs directed against cytokine receptors and that have triggered cytokine release syndrome when evaluated in mouse xenograft models, owing to target engagement in healthy tissues.10 Nonetheless, the data presented suggest that TSLPR represents a promising immunotherapeutic target in a high-risk and poor-prognosis subset of B-ALL, and therefore merits further clinical development.
Conflict-of-interest disclosure: The authors declare no competing financial interests.