Acute Myeloid Leukemia (AML) Blasts Influence the Gene Expression Signature and Co-Signaling Receptor Expression of CD8⁺ T Cells

Background: T cell dysfunction in AML remains poorly understood. Our previous studies of AML-associated T cell dysfunction (Knaus, ASH 2015) have focused on expression of multiple inhibitory receptors by T cells in AML patients. Transcriptional signatures, however, remain relatively unexplored, as does the role of Blast/T cell interactions on T cell function. Deciphering those could be crucial for integration of future immunotherapies into clinical practice. Therefore, we aimed to characterize CD8⁺ T cell gene expression signatures in newly diagnosed AML patients before and after treatment, and to decipher the effects of AML blasts on the expression of co-signaling molecules by CD8⁺ T cells in co-culture experiments.

Methods: Serial peripheral blood (PB) samples (at diagnosis and at the recovery after induction chemotherapy) were collected. To study transcriptional signatures, RNA isolated from FACS-purified PB CD8⁺ T cells from 6 patients [3 responders (R) and 3 non-responders (NR)] and 4 healthy controls (HC) was analyzed with the Human Prime View Gene Expression Array (Affymetrix). The data were normalized and log transformed. Expression fold change (FC), p values and false discovery rate were determined. Enrichment of canonical pathways was determined using Ingenuity Pathway Analysis (IPA, QIAGEN).

To study AML blast-T cell interactions, we FACS-purified T cells and primary AML blasts at diagnosis (n=13) and T cells from HC (n=12). T cells were cultured in vitro for 3 days in the presence or absence of blasts (T cell:blast ratio 1:10) and analyzed by flow cytometry.

Results: The transcriptional profile of CD8⁺ T cells at AML diagnosis significantly differed from that of HC. Genes were selected based on >2 FC between patient and HC, and p< 0.01. We identified a total of 453 dysregulated genes, of which 237 were up- and 216 down-regulated. Upregulated genes included immune inhibitory receptors LILRB1, 2B4, KLRG1, CD160, the transcription factors EOMES, TBET, TIGIT and cytokines (granzyme-A/B/K). In contrast, co-stimulatory receptor genes were downregulated, including CD40LG, CD28, CD30LG and CD28H. Canonical pathways analysis with IPA revealed that the NFAT pathway (involved in T cell differentiation and self-tolerance) was highly upregulated, while co-stimulatory CD28, ICOS and OX40 signaling pathways were downregulated in CD8⁺ T cells at AML diagnosis.

Next, we compared R to NR after induction chemotherapy. There were a total of 351 dysregulated genes; 108/243 genes were up-/down-regulated, respectively. R patients upregulated immune stimulatory receptor genes like ICOS, whereas the top expressed genes for NR patients included the co-inhibitory receptor TIM3; several members of the inhibitory LIR receptor family; LST1 (involved in inhibition of lymphocyte proliferation); TWEAK-APRIL (associated with T cell apoptosis); and CD39 (terminally exhausted CD8⁺ T cells). In line with these findings, IPA showed that the co-stimulatory ICOS and OX40 signaling pathways were enriched in R patients. In contrast, the NFAT pathway, which had been highly upregulated at diagnosis, remained enriched in NR, but not in R patients. Results were confirmed by qPCR.

The culture assay showed that the presence of primary AML blasts significantly reduced the viability of both AML and HC T cells (p <0.005 in both cases). The presence of AML blasts also significantly decreased the frequency of primary AML T cells expressing co-stimulatory receptors 41BB, ICOS and OX40, while it increased the frequency of HC T cells expressing co-inhibitory receptor 2B4 and the senescence/exhaustion marker CD57 compared to their counterparts cultured without blasts.

Conclusions: Our study provides insight into the genomic CD8⁺ T cell signatures of AML patients at diagnosis and following chemotherapy. At diagnosis, T cells overexpressed genes that negatively regulate T cell immune responses, while genes that positively regulate immune responses were downregulated. Interestingly, after induction chemotherapy these changes persisted in NR only. Additionally, a pattern of decreased viability and co-stimulatory receptor expression was seen after in vitro co-culture of T cells with AML blasts, whereas immune inhibitory receptor expression was increased. Our data suggests that the blasts themselves influence the T cell phenotype and genotype in AML patients and that remission is associated with reversion to HC pattern.