Noncanonical Function of AGO2 Augments T-cell Receptor Signaling in T-cell Prolymphocytic Leukemia

T-cell prolymphocytic leukemia (T-PLL) is a rare peripheral T-cell neoplasm, yet it represents the most common form of mature T-cell leukemias in the United States and Europe. It is characterized by an aggressive and chemotherapy-refractory course with currently very limited therapeutic options (1, 2). The median overall survival of patients is 20–36 months (3). Patients with T-PLL typically present with exponentially rising lymphocyte counts and with bone marrow infiltration accompanied by marked (hepato)splenomegaly. Patients do not respond well to standard multiagent chemotherapies. Long-term remissions of >5 years can be accomplished in <20% of patients, but only if they undergo an allogeneic stem cell transplantation in first remission (2, 4). Such otherwise short-lived remissions are usually induced by the anti-CD52 anitibody alemtuzumab.

T-PLL cells are postthymic, antigen-experienced T lymphocytes (5, 6). Expression and functional competence of the T-cell receptor (TCR) is found in 80% of T-PLL cases, which is associated with a poor-prognostic hyperproliferative phenotype (6, 7). Moreover, global gene expression signatures in T-PLL resemble those of T-cell activation, further pointing toward a prominent role of TCR signals in T-PLL pathophysiology (8). However, despite these advances, novel rational treatment designs are urgently needed and these would tremendously benefit from a refined understanding of T-PLL's biology.

As one molecular hallmark and a central initiating lesion, >90% of T-PLL harbor chromosomal aberrations, that is inv14 or t(14;14), which lead to constitutive expression of the proto-oncogene T-cell leukemia 1A (TCL1A). This adapter molecule amplifies TCR signals via enhanced kinase activation in leukemic precursors and in tumor cells of T-PLL. In a threshold-lowering fashion, the aberrant TCL1A expression allows the (pre)leukemic memory-type T cell to drive on low-level (autonomous) TCR input and both signals, by TCL1A and by engaged TCR, cooperate toward T-PLL development (6–8). Additional recurrent genomic alterations in T-PLL target the ataxia telangiectasia–mutated tumor suppressor gene through deletions and damaging mutations as well as the STAT5B/JAK1/JAK3 genes through gain-of-function mutations (8, 9).

In genome-wide analyses of somatic copy-number alterations (CNA) we previously identified gains on chromosome 8q in approximately 30% of T-PLL cases. Importantly, we discovered that these always (100%) involve the locus of argonaute RISC catalytic component 2 (AGO2), and at lower frequencies (∼70%) the MYC proto-oncogene and the miRNAs (miR)-1206/1207/1208 (8). Recent descriptions of miR profiles in T-PLL revealed resemblance with those of TCR-activated T cells and integration with transcriptome data highlight regulatory networks of survival signaling and DNA damage responses, but the underlying causes of these changes in the leukemic miR-ome remained to be uncovered (10).

miRs assume crucial roles in cancer initiation and progression. Their deregulated levels are mostly attributable to genetic or epigenetic alterations that affect components of the miR processing machinery, such as AGO2, DROSHA, XPO5, or DICER (11). AGO2 encodes for a component of the RNA-induced silencing complex (RISC) that is responsible for the inhibition of mRNA transcription upon miR binding and is, therefore, classified as the central mediator of miR processing (12). Among the four mammalian argonaute proteins, AGO2 is the only member that is endowed with a catalytic domain. Its endonuclease activity is essential for cleavage of the miR/siRNA passenger strand as well as of the RNA that is targeted by the miRs/siRNAs.

In the current study, we demonstrate a generally elevated AGO2 expression in primary T-PLL samples. Highest levels were associated with more aggressive disease parameters such as tumor burden, with a proproliferative cell phenotype, and with specific changes in miR-omes and transcriptomes. Further characterizing the potential oncogenic function of AGO2 in T-PLL, we describe AGO2 interactomes in the context of TCR activation and discover protein–protein interactions of AGO2, that is, with the ζ-chain–associated protein kinase 70 (ZAP70) that, at least in part, underlie the AGO2-dependent TCR hyperactivation in T-PLL.

Peripheral blood (PB)-derived T-cell isolates were obtained from 56 patients with T-PLL and from 8 age-matched healthy donors. The majority of samples (n = 47/56; 83.9%) were from treatment-naïve patients. The median age of the cohort was 68 years (range from 32–88 years) and 48.2% of patients were females (summarized data in Supplementary Table S1).

Details on isolation of PB mononuclear cells and magnetic bead–based cell enrichment of pan-CD3⁺ T cells are presented in the Supplementary Materials and Methods. Patients with T-PLL fulfilling the consensus criteria for the diagnosis of T-PLL (2) were included in the study. We did not define exclusion criteria for this study and patients were not randomized into groups because this was deemed irrelevant to this study. Written informed consent according to the Declaration of Helsinki was provided by all patients and healthy donors. Collection and use of the samples were approved for research purposes by the ethics committee of the University Hospital of Cologne (Cologne, Germany; #12-146, #19-1085, and #19-1089). Individual demographics and detailed information on clinical characteristics, cytogenetics, immunophenotypes, and follow-up for all 56 patients are presented in Supplementary Table S2).

For all patients of this 56-case set, triplicate immunoblot studies for AGO2 protein expression were performed following standard protocols (see Supplementary Materials and Methods; AGO2 antibody clone 11A9, gift of G. Meister, University of Regensburg, Regensburg, Germany). To assess for clinical associations with AGO2 expression, patients were divided in tertiles based on their AGO2 protein levels. Overlap of all profiling data (Supplementary Fig. S1A) allowed associations of AGO2 protein with AGO2 mRNA and with AGO2 copy number (CN) status in 46 of these 56 patients.

Moreover, we previously performed transcriptome and small RNA sequencing in a cohort of 46 T-PLL (10). Details on RNA isolation, sequencing, and data processing are given in (10). As part of this study, we obtained AGO2 protein expression data in 41 of those 46 patients with T-PLL. Differentially expressed genes and miRs between the high AGO2 protein tertile (n = 14) versus low AGO2 tertile (n = 14) were calculated (FDR < 0.05). Gene set enrichment analyses (GSEA) were performed on differentially expressed mRNAs as well as on sorted lists of protein-coding genes, based on Spearman correlation coefficients for each differentially expressed miR. Hallmark gene sets were applied (13). The Mann–Whitney–Wilcoxon (MWW) test was performed for systematic comparison of continuous data and the Fisher exact test for categorical data. A P value <0.05 was considered statistically significant.

Primary T-PLL cells were cultured in suspension at concentrations of 2 × 10⁶ cells/mL in RPMI1640 (Gibco, catalog no. 21875-034), supplemented with 10% FBS (Gibco, catalog no. 10270-106) and penicillin/streptomycin at 100 U/L (Gibco, catalog no. 15140-122). The T-cell leukemia/lymphoma cell lines Jurkat (male-derived, ATCC, catalog no. TIB-152, RRID:CVCL_0367) and the male-derived Hut78 (ATCC, catalog no. TIB-161, RRID:CVCL_0337) were used for in-vitro experiments. The identity of all cell lines was recently authenticated by short tandem repeat analysis. Presence of Mycoplasma was periodically tested by PCR. Their suspensions (0.5–1.0 × 10⁶ cells/mL) were maintained in RPMI1640 with 10% FBS. Details on transient AGO2 knockdowns, TCR stimulation, induction of DNA damage, and on assessment of cell viability, proliferation, and survival are presented in the Supplementary Materials and Methods. FISH, qRT-PCR, and immunoblots were performed according to standard procedures (8) and described in the Supplementary Materials and Methods.

Phosphorylation of 43 human protein kinases in relation to two reference proteins was measured by an immunoblot-based human phosphokinase array (R&D Systems, catalog no. ARY003B). Details are given in the Supplementary Materials and Methods.

Detailed protocols on the analysis of the composition of immuno-isolated protein complexes are provided in the Supplementary Materials and Methods. Generally, precleared protein lysates of three technical replicates each were incubated with bead-bound protein G–coupled AGO2 antibodies (Abcam, catalog no. ab186733, RRID:AB_2713978) or IgG isotype control (Cell Signaling Technology, catalog no. 2975, RRID:AB_10699151). Proteins eluted from the beads were denatured and electrophoretically separated. After trypsin and endoproteinase Lys-C based in-gel digestion and extraction, resulting peptides were subjected to high-resolution mass spectrometry (MS) in the CECAD Proteomics facility (for protocols, equipment, and raw data processing algorithms see Supplementary Materials and Methods).

Jurkat or primary T-PLL cells, untreated or pretreated with 10 µmol/L of the LCK inhibitor CAS 213743-31-8 (Sigma-Aldrich, catalog no. 428205-M), or with 10 µmol/L dasatinib (MedChemEx, HY-10181, CAS 302962-49-8) for 24 hours where indicated, were stimulated on anti-CD3 (BioLegend, catalog no. 317348 clone OKT3, RRID:AB_2571995)-coated coverslips for indicated times and fixed with paraformaldehyde. After blocking of the cross-linking antibody by incubation with IRDye800 anti-mouse antibody (LI-COR Biosciences, catalog no. 926-32212, RRID:AB_621847) for 1 hour at room temperature and additional paraformaldehyde fixation for 10 minutes, cells were permeabilized by 0.5% Triton X-100 in PBS for 5 minutes. Unspecific antibody binding was blocked for 15 minutes by 1% BSA in PBS, before samples were incubated with primary antibodies against ZAP70 (Cell Signaling Technology, catalog no. 3165, RRID:AB_2218656, 1:100) and AGO2 (Santa Cruz Biotechnology, catalog no. sc-53521, RRID: AB_628697, 1:50) for 1 hour at room temperature in 1% BSA in PBS.

For confocal microscopy, samples were incubated with anti-mouse Cy3 (Jackson ImmunoResearch, catalog no. 715-165-150, RRID:AB_2340813) and anti-rabbit alexa488 (Jackson ImmunoResearch, catalog no. 711-545-152, RRID:AB_2313584) secondary antibodies for 1 hour at room temperature. For proximity ligation assays (PLA), the Duolink In Situ Red Starter Kit Mouse/Rabbit (Sigma-Aldrich, catalog no. DUO92101) was applied according to manufacturer's instructions starting with the secondary antibodies. Nuclei were visualized using DAPI (Sigma, catalog no. D9542). Images were acquired on a Leica SP8X with white laser and HyD detector using a 63× objective and standard settings for sequential imaging. For data analyses, the FIJI/imageJ software was used to quantify PLA spots. For each condition, 50–100 cells were analyzed. Statistical significance was determined by a Mann–Whitney rank-sum test.

Singular models: Experimentally determined structures were integrated with predicted models (14) to gain highest possible accuracy of the implicated interactions. Recompiled information on the phosphorylation states of active forms was taken from UniProt (15) to amend the models accordingly.

Complexes of AGO2 with ZAP70 and ZAP70 with LCK: Searches of productive interactions by in silico docking with ambiguous interaction restraints defined around putative targets [relied on comparative alignments and data accessible on the PhosphositePlus (16)] were completed using the program Haddock (17) on the experimentally determined structures of the considered domains. The resulting models, in which the targeted tyrosine was too far away from the gamma phosphate of ATP, were considered unlikely based on the assumption that the interaction is to result in tyrosine phosphorylation. Each chain of retained models was fused back into its relative structure and the whole complex was further challenged in a 60 ns molecular dynamics run in a water box.

Functional model TCR:LCK:ZAP70:AGO-2: The conjectural model was based on the motif analyses and SH2 and SH3 domain specificities. The placement of each protagonist was deduced from accessibilities and executed “by hand” in a PyMOL session. The whole system including a POPC membrane was built-up using the Membrane Builder module from CHARMM-GUI (18).

Short molecular dynamics or minimizations: The complex models were used to build systems in all-atoms molecular dynamics in explicit water at 150 mmol/L KCl with CHARMM-GUI (18). GROMACS (19) was used for these tasks. The Zap70:AGO-2 complex was challenged for 60 ns in a system of 464,189 atoms, while the conjectural model embedded in a membrane and consisting in a system of 1,778,446 atoms was simulated for a 10 ns, both remaining steady in the intervals. Simulations were executed by keeping the temperature at 310.15 K and pressure at 1 atm. Standard CHARMM-GUI output protocols were used.

MS-based proteomics data are deposited at the ProteomeXchange Consortium via the PRIDE (20) partner repository with the dataset identifier PXD028170. Details on data processing and data availability of total and small RNA expression data used in this study are given in ref. 10. Further information and requests for resources and reagents should be directed to and will be fulfilled by the corresponding author.

Central piece of this study was an integrated multilevel dataset derived from a cohort of 46 well-annotated T-PLL cases. Their primary PB-derived T-PLL cells were analyzed for CNAs, miR-ome profiles, and transcriptome signatures (Supplementary Fig. S1A). AGO2 protein expression was analyzed in 41 of those and in 15 additional cases (Supplementary Table S2). At the outset, we recapitulated our SNP-array based observation that genetic amplification of AGO2 was involved in all of those T-PLL cases that harbored a minimally amplified region (MAR) on chromosome 8q (8). Here, all cases harboring this MAR showed a gain of AGO2 (n = 12/12, average CN = 2.4; Supplementary Table S1), while the neighbouring MYC locus was only involved in approximately 80% of cases (n = 10/12, average CN = 2.3). Figure 1A illustrates biallelic MYC adjacent to an AGO2 CN gain. Genomic losses of AGO2 were not discovered here, neither were they reported previously (8, 21). Corresponding with this overrepresentation of AGO2, we observed elevated expression of AGO2 mRNA in eight of nine isolates from primary T-PLL samples (by qRT-PCR) and of AGO2 protein (immunoblot-based levels) in cells from 41 of 56 analyzed cases as compared with healthy donor–derived pan-CD3⁺ T cells [fold-change (fc) >1.5; Fig. 1B–D; Supplementary Tables S1 and S2]. While AGO2 mRNA expression significantly correlated with the corresponding protein expression, there was no significant correlation between AGO2 CN and protein expression, indicating additional ways of AGO2 upregulation beyond the genomic gains (Fig. 1E; Supplementary Fig. S1B and S1C). T-PLL with high AGO2 protein expression had a mean AGO2 CN of 2.43 (vs. mean AGO2 CN of 2.27 in those with low AGO2 protein expression).

We next associated AGO2 expression with clinical characteristics. Higher AGO2 protein levels were predominantly found in cases with more aggressive features, as represented by higher blood lymphocytosis, a shorter lymphocyte doubling time, more reduced platelet counts, and higher serum lactate dehydrogenase (LDH) levels (Fig. 1F and G; Supplementary Fig. S2A–S2D; Supplementary Table S3). Subsequent AGO2 knockdown experiments in Jurkat T cells and in primary T-PLL samples underlined the proactive tumor-promoting role of AGO2. Induced reduction of AGO2 negatively affected cellular metabolic activity [as measured by NAD(P)H flux], viability, and proliferative capacity (Fig. 1H–J; Supplementary Fig. S3A and S3B). Negative CERES dependency scores in hematopoietic and T-cell leukemia/lymphoma lines indicated that AGO2 is essential for most represented lines (depmap portal; Supplementary Fig. S3C).

To dissect how AGO2 might mediate this proleukemic effect, we followed two distinct strategies: (i) to evaluate global miR-mediated protumorigenic functions of AGO2 by analyzing miR/mRNA expression profiles associated with high AGO2 levels and (ii) to define immediate, potentially miR-independent, effects of high-level AGO2 on T-PLL–intrinsic oncogenic pathways through identification of direct AGO2–protein interactions.

Thus, we first analyzed in an integrated fashion the transcriptomes and miR-omes of primary T-PLL (generated as part of ref. 10) in association with stratified AGO2 protein expression. Using upper and lower tertiles, we compared n = 14 “AGO2-high” versus n = 14 “AGO2-low” cases (see Materials and Methods and Supplementary Table S2 for definition) with respect to AGO2-associated miR/mRNA deregulations (Fig. 2A and B). We identified an overall higher number of downregulated mRNAs (n = 209 vs. n = 30) and miRs (n = 62 vs. n = 5) in AGO2-high cases, suggesting a globally enhanced miR-mediated mRNA degradation (Supplementary Fig. S4A and S4B; Supplementary Table S4). This is in line with AGO2’s established central function in RISC-mediated gene silencing.

miR-mediated regulation of cellular processes is based on the induction of low-level changes of mRNA abundances affecting a larger cohort of genes rather than one single factor (10). Thus, we refrained from analyzing specific single genes in detail but conducted GSEAs using the complete list of AGO2-associated mRNAs and miRs (Supplementary Table S4). As GSEA can only be performed on mRNA expression data (not miRs), we performed the latter analyses on mRNAs predicted to be regulated (and thus differentially expressed) by the identified AGO2-associated miRs.

Particularly mRNAs and miRs associated with the GSEA-based Hallmark pathways of survival signaling (e.g., “IL2/STAT5 signaling”), cell-cycle control (e.g., “E2F targets” or “G₂–M Checkpoint”), and DNA damage responses (e.g., “DNA repair” or “p53 Pathway”) were significantly affected (T-PLL with high vs. low AGO2 protein; Fig. 2C; Supplementary Fig. S4C; Supplementary Table S4). Similar associations of AGO2 expression with gene expression profiles were found in cell line systems of short hairpin RNA–mediated AGO2 knockdown (Supplementary Fig. S5A and S5B; ref. 22), pointing toward a potential regulatory involvement of AGO2 in these processes.

Besides its role in global miR/mRNA network regulation, AGO2 has been described to also directly interact with and to impact factors involved in pro-oncogenic pathways, for example, RAD51 in DNA damage responses (23). Moreover, given the central role of the TCR pathway in growth and survival of the T-PLL cell (6, 7) and to assess for nonconventional functions of AGO2, that is, in regulating intermediates of the TCR cascade, we interrogated the impact of modulated AGO2 on TCR signaling. First, using antibody-based phosphokinase arrays, we observed diminished TCR-induced responses (anti-CD3 cross-linking) subsequent to siRNA-mediated AGO2 knockdowns in Jurkat T cells (selected kinases in Fig. 3A, examples in Supplementary Fig. S6A–S6C, for example, TOR, fc = 0.38; Src, fc = 0.47). While AGO2 knockdown cells presented a strong upregulation of p27 in the unstimulated condition (fc = 7.34), we showed downregulation of TCR-associated kinases in both conditions upon AGO2 downregulation (e.g., unstimulated condition: TOR, fc = 0.60; Src, fc = 0.58; stimulated condition: TOR, fc = 0.38; Src, fc = 0.47). Next, the positive effect of AGO2 on TCR-triggered kinase activation was specifically validated for pZAP70, pPLCγ1, and pLAT in separate AGO2 knockdown experiments in Jurkat cells (Fig. 3B; Supplementary Fig. S7A). Notably, we did not observe an altered expression or signal-induced phosphorylation of the proximal TCR kinase LCK upon AGO2 knockdown (Supplementary Fig. S7B). We further validated the enhancing effect of AGO2 upregulation on the TCR signaling cascade downstream of LCK in the Hut78 T-cell leukemia line (Supplementary Fig. S7C) as well as in AGO2-level stratified primary T-PLL samples (Fig. 3C). As for AGO2 protein expression itself, we observed elevated levels upon TCR stimulation in pan-CD3⁺ T cells of one out of healthy donor samples and in 1 patient with T-PLL, with a stronger basal and longer-lasting upregulation of AGO2 in the T-PLL case (Supplementary Fig. S8).

To control for the possibility of TCR-induced signaling responses being altered merely due to an impaired miR processing machinery, we experimentally modified DICER expression levels in Jurkat cell systems to induce a global reduction of RISC complex function (Supplementary Fig. S9A–S9C; ref. 24). Detecting phosphorylation of LAT, ZAP70, and PLCγ1 upon TCR activation via anti-CD3 cross-linking and using immunoblots, we did not identify alterations in TCR pathway responses in DICER knockdown conditions.

These AGO2 knockdown systems also allowed us to validate the role of AGO2 in regulating its “conventional” effectors, such as the tumor suppressor p27. AGO2-depleted Jurkat T cells showed a strong upregulation of p27 in the TCR-unstimulated baseline condition (Supplementary Fig. S6C, fc = 7.34). According to p27’s cell cycle and survival-regulating functions, this was associated with a reduced apoptotic resistance in response to irradiation-induced DNA damage (Supplementary Fig. S10).

To interrogate the TCR-enhancing impact of AGO2 as RNA-independent effects of the AGO2 protein in more detail, that is, to identify the underlying molecular mediators, we conducted immunoprecipitation experiments of endogeneous AGO2 in TCR-unstimulated versus TCR-activated Jurkat T cells (naturally TCL1A negative; Fig. 4A). Short-time TCR stimulation of 5 minutes was chosen to minimize miR-based and transcriptional effects on the interactome of AGO2. Subsequent MS analysis of AGO2-immune complexes identified 698 AGO2-interacting proteins common to both experimental conditions (Fig. 4B–D; Supplementary Fig. S11A and S11B; Supplementary Table S5), while 44 proteins were specifically enriched in the TCR-activated condition (Supplementary Fig. S11C). A functional network analysis of significantly enriched Reactome pathways (P ≤ 0.001) showed a dominant association of AGO2 interactors with pathways of “RNA metabolism” (Fig. 4E; e.g., survival of motor neuron SMN1). This also identified coprecipitated proteins to be involved in regulation of “gene expression” (e.g., RISC-loading complex subunit TARBP2, histone deacetylase HDAC1), “DNA repair” (e.g., PARP1, an ADP-ribosylating enzyme essential for initiating various forms of DNA repair), and “cell-cycle control” (e.g., the cyclin-dependent kinase CDK1).

Importantly, the analysis of AGO2-antibody enriched proteins additionally identified LCK and ZAP70, both central kinases of the TCR signal cascade, as AGO2 protein complex constituents. We validated the interaction of AGO2 with ZAP70 and with phospho-ZAP70^Tyr319 in Western blot analyses of AGO2 immunoprecipitations in Jurkat T cells (Fig. 4F) and in primary T-PLL cells (n = 3, Fig. 4G). Of note, these protein interactions were predominantly detected under TCR-stimulated conditions.

To validate and to further interrogate the TCR-induced AGO2-ZAP70 complex formation, we applied confocal microscopy and PLAs in Jurkat cell line systems and primary T-PLL cells (Fig. 5; Supplementary Fig. S12). The amount of induced cytoplasmic membrane–associated AGO2-ZAP70 overlap peaked at 10 minutes subsequent to anti–CD3-mediated TCR activation as detected by AGO2 (green) and ZAP70 (red) fluorescence signals (Fig. 5A) and by PLA-assessed spots per cell (Supplementary Fig. S12A). TCR-signal dependence of AGO2-ZAP70 complexes was further underlined by inhibition of LCK through pretreatment of cells with CAS213743-31-8 or dasatinib (Fig. 5B and C; Supplementary Fig. S12B and S12C). Such blocking of LCK-mediated signal transduction in anti–CD3-stimulated Jurkat cells significantly reduced AGO2-ZAP70 complex formation to levels that were similar to those of unstimulated conditions (Fig. 5B). These observations could be recapitulated in primary T-PLL cells of two independent cases (Fig. 5C; Supplementary Fig. S12D).

To resolve and model the organization of the AGO2-ZAP70 complexes, we undertook a series of dockings with Haddock (targeting different tyrosines likely to undergo this modification) using ambiguous constraints around the target and involving the kinase mouth. A conclusive result for tyrosine 529 phosphorylation was retained [HADDOCK score—45.2 ± 11.0 (17)] (Fig. 6A–C; ZAP70, yellow; AGO2, pink) and placed in the structural architecture of the two partners further modeled in full-length versions. The full-length model (Supplementary Fig. S13A and S13B) was relaxed by a 40 ns molecular dynamics run [with GROMACS (19) in an all atoms system prepared with CHARMM-GUI (18) and typed with CHARMM36 m force field (25)]. This constitutes per se a possible modification model of a yet characterized phosphorylation site for AGO2 that supports ZAP70 to be a direct modulator of AGO2 activity. The results from our coimmunoprecipitation screening (Fig. 4) and functional colocalization assays (Fig. 5) point toward involvement of the TCR-central kinase LCK in the AGO2-ZAP70 complex formation. To contextualize this more and given suggested additional specific interactions at a priori complementary elements of structure and guided by existing similar complexes, we undertook the arrangement of a putative interaction of LCK with AGO2-ZAP70 at ITAMs of the TCR-associated CD3ζ chain (Fig. 6D; Supplementary Fig. S14A). Figure 6D shows a full view of the complex highlighting different interactors and key points for interrecognition. The presented model favors an LCK-mediated preactivation of ZAP70 by phosphorylation of its Y319 (Supplementary Fig. S14B) followed by phosphorylation of Y529 of AGO2 by ZAP70 (Supplementary Fig. S14C). Overall, our model suggests enhanced stability of LCK-AGO2-ZAP70 complexes through predicted protein–protein interactions, potentially explaining the TCR signal–enhancing effect of high AGO2 levels.

Overall, we describe that the frequent AGO2 overrepresentations at the genomic level underlie, but do not entirely explain, a general AGO2 protein overexpression in T-PLL. Higher AGO2 levels seem to promote T-cell leukemia growth. Besides AGO2-associated global impacts on miR and transcriptome profiles with expected implications in cell survival, cell-cycle control, and DNA damage responses, we also discover noncanonical AGO2 protein–mediated effects beyond RNA metabolism. Particularly the activating interaction of AGO2 with TCR kinases indicates that the strength of protumorigenic TCR signals can be determined by AGO2 levels.

The mechanisms of AGO2 in small-RNA guided gene silencing are diverse and up to now not fully understood (12). On the basis of our findings of AGO2 gene amplifications and protein overexpressions in T-PLL, we pursued the potential oncogenic role(s) of AGO2 in the pathogenesis of this T-cell malignancy. We characterized global AGO2-shaped miR-omes and transcriptomes in T-PLL and highlight that especially pathways of survival signaling, cell-cycle control, and DNA damage responses are affected in a proleukemic fashion by higher AGO2 levels.

In malignant B cells, miRs interconnect damage signals upstream of p53 and B-cell receptor signal strength, for example, through FOXP1 or phosphatase levels (26). Importantly, we describe for the first time a functional link of elevated AGO2 protein with enhanced TCR signaling activity at the level of kinase enhancements. Generally, regulatory impacts of several miR species on TCR signaling in T-cell physiology have been described. AGO2 itself can be subject to TCR-triggered proteasomal degradation (27, 28). Interestingly, we noted increased AGO2 protein levels upon TCR stimulation in pan-CD3⁺ T cells of one healthy donor and 1 patient with T-PLL. Although we observed an association of AGO2 CN with mRNA expression in T-PLL cells, AGO2 protein expression levels appeared less correlated with AGO2 CN. This indicates further modes of AGO2 upregulation in T-PLL beyond genomic aberrations, which have been observed already in other cell systems, for example, regulation of AGO2 protein stability by miR availability (29).

In addition to miR-mediated transcriptome modifications, direct protein–protein interactions with specific signaling intermediates as well as posttranslational modifications of AGO2 induced by upstream kinases have been shown to contribute to AGO2’s oncogenic function (12, 30). The study presented here, demonstrates a previously undisclosed relationship of AGO2 with TCR signaling, showing that (i) AGO2 augments TCR signal strength through protein–protein interactions and that (ii) TCR activation influences the AGO2 interactome, both eventually additionally affecting miR/transcriptome profiles.

With data on RNA-mediated and noncanonical direct protein-related effects of AGO2 in TCR signaling, our study adds to the understanding of AGO2’s likely wide spectrum of functions, generally and in the context of T-cell tumorigenesis. Antibody-based phosphokinase arrays implicated further functions of AGO2 besides its involvement in TCR signaling (e.g., p27 deregulation), which should be task of future research.

The biochemical specifics of the interactions of the AGO2 protein with LCK and ZAP70 require more detailed analyses. We conclude that T-PLL cells are equipped with a miR and transcriptional network that is shaped by a cooperative impact of TCR signals and those by AGO2, which enhance each other at the protein level. The observations we made here entail the future challenge of distinguishing the context specifics between (conventional) “indirect” miR-mediated versus (noncanonical) “direct” RNAi-independent pro-oncogenic effects of AGO2. The immediate nature of its recruitment to protein complexes within minutes subsequent to TCR activation, argues in favor of a miR-independent involvement of AGO2 in the regulation of TCR signals. However, to assess miR-mediated long-term effects (additionally) shaping enhanced TCR-derived signals, the basal miR processing machinery has to be modified in an experimental setting. This would cause an intrinsically altered artificial cell that is not comparable with our system of interest (31). Further studies also need to address the predicted impact of TCR kinases (i.e., ZAP70) on AGO2 activity (see model in Fig. 6D). Phosphorylation of AGO2 at its residue Y529, predicted to be essential for the binding of ZAP70, was shown to be associated with strongly reduced small RNA binding, further pointing toward a noncanonical function of AGO2, when phosphorylated at Y529 (32).

Overall, our findings unravel a nonconventional mode of action of AGO2, namely via protein–protein interactions in the context of TCR signals, in addition to its classical miR/mRNA-mediated effects. This also further underlines the central role of TCR signaling in T-PLL's pathophysiology. Besides TCL1A as a catalytic enhancer of TCR kinases (i.e., ERK and AKT; refs. 6–8) and besides loss of negative regulators of T-cell activation (i.e., CTLA4 and other coreceptors; refs. 6, 8), the activating effects of the AGO2 protein on ZAP70, PLCγ1, and LAT described here, represent a new mode of promoting a TCR-hyperactivated phenotype of the T-PLL cell or its precursor. This emphasizes TCR signaling intermediates as candidates for targeted therapeutic approaches in T-PLL, for example, as pursued by first generations of IL2-inducible T-cell kinase inhibitors (33).

No disclosures were reported.

T. Braun: Conceptualization, resources, data curation, formal analysis, validation, investigation, methodology, writing–original draft. J. Stachelscheid: Conceptualization, data curation, formal analysis, investigation, methodology, sample acquisition. N. Bley: Data curation, formal analysis, investigation, methodology. S. Oberbeck: Data curation, formal analysis, investigation. M. Otte: Data curation, formal analysis, investigation. T.A. Müller: Data curation, formal analysis, investigation, methodology. L. Wahnschaffe: Data curation, formal analysis, methodology. M. Glaß: Data curation, formal analysis, methodology. K. Ommer: Resources, data curation, sample acquisition. M. Franitza: Resources, data curation, supervision. B. Gathof: Resources, data curation, supervision. J. Altmüller: Resources, data curation, supervision. M. Hallek: Resources, data curation, supervision, project administration, writing–review and editing. D. Auguin: Resources, data curation, supervision, funding acquisition, writing–original draft, project administration, in silico modeling. S. Hüttelmaier: Resources, supervision, funding acquisition, project administration, writing–review and editing. A. Schrader: Resources, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, project administration. M. Herling: Resources, supervision, funding acquisition, project administration, writing–review and editing.

This research was funded by the DFG Research Unit FOR1961 (Control-T; HE3553/4-2), the Köln Fortune program, and the Fritz Thyssen Foundation (10.15.2.034MN). This work was also funded by the EU Transcan-2 consortium “ERANET-PLL” and by the ERAPerMed Consortium “JAKSTAT-TARGET.” A. Schrader was supported by a scholarship of the German José Carreras Leukemia Foundation (DJCLS 03 F/2016). MS analyses were conducted at the CECAD proteomics facility. The authors gratefully acknowledge their help during data acquisition and data analysis. Furthermore, the authors thank the ENCODE Consortium and the Brenton Gravely Lab for making AGO2 knockdown data available. Finally, they thank all patients with their families for their invaluable contributions.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.