Abstract

   Peripheral T cell lymphoma (PTCL) is a heterogeneous group of
   postthymic T cell neoplasms, with ~40% classified as PTCL–not otherwise
   specified (PTCL-NOS). PTCL-GATA3, a molecularly defined subtype,
   associated with T helper 2 (T[H]2)–like differentiation and poor
   prognosis, has frequent co-occurrence of TP53 loss/mutation and
   heterozygous PTEN loss. CD4+ T cell conditional mouse models with Trp53
   mutation/deletion and Pten loss demonstrated mature T cell lymphomas
   (mTCLs) with T[H]2-like transcriptomic and immunophenotypic profiles.
   Molecular studies revealed that codeletion of Trp53/Pten induced T cell
   receptor and Janus kinase–signal transducer and activator of
   transcription signaling, promoting T[H]2 differentiation while
   inhibiting T[H]1 differentiation. These findings were validated by
   CRISPR editing of TP53/PTEN loss in human CD4+ T cells and
   mechanistically evaluated the p53 binding region in intron-3 of GATA3,
   resulting in transcriptional repression. Transcriptomic profiles of
   m-TCLs recapitulated human-PTCL-GATA3 transcriptome and distinguished
   PTCL-NOS subtypes. Preclinical assessment of m-TCLs with PI3Kγ/δ
   inhibitors significantly improved survival, supporting a therapeutic
   approach for the p53-aberrant PTCL-GATA3.
     __________________________________________________________________

   TP53 and PTEN loss drives T[H]2 differentiation and T cell
   lymphomagenesis, highlighting increased sensitivity to PI3K inhibitors.

INTRODUCTION

   Peripheral T cell lymphoma (PTCL) is a heterogeneous group of mature T
   cell malignancies ([84]1, [85]2), and patients with PTCL have inferior
   clinical outcomes compared with patients with B cell lymphoma, with a
   5-year overall survival (OS) of <30% using current therapeutic regimens
   ([86]3). These lymphomas are difficult to diagnose, requiring a
   multiparameter approach that integrates clinical history, morphology,
   immunophenotypic findings, and genetic analysis to ensure an accurate
   diagnosis ([87]2, [88]4). About 30 to 50% of PTCLs are defined as
   PTCL-not otherwise specified (NOS) through diagnostic exclusion of the
   other World Health Organization (WHO)/International Consensus
   Classification (ICC) specified entities ([89]1, [90]2). Using
   multiomics approaches, we identified two major distinct biological and
   prognostic subtypes within PTCL-NOS ([91]5, [92]6) that are indicated
   in the current WHO/ICC classification of lymphoid neoplasms ([93]7).
   Thirty-five percent of PTCL-NOS cases are characterized by high GATA3
   expression and its transcriptional target genes, designated as the
   PTCL-GATA3 subtype, and ~55% have high expression of TBX21 and its
   target genes, designated as the PTCL-TBX21 subtype. GATA3 and TBX21 are
   the master transcriptional regulators of T helper 2 (T[H]2) and T[H]1
   cell differentiation, respectively, and regulate the expression of
   relevant cytokines ([94]8, [95]9). The biological and prognostic
   differences between PTCL-GATA3 and PTCL-TBX21 subtypes have been
   validated in other previous studies ([96]10, [97]11).

   PTCL-GATA3 and PTCL-TBX21 subtypes demonstrated distinctive oncogenic
   pathways, and the PTCL-GATA3 subtype showed recurrent aberrations in
   the p53 and PI3K pathways due to frequent del17p13 (TP53) and del10p23
   (PTEN) ([98]12). Whereas del17p was associated with TP53 mutation or
   biallelic loss, del10p23 was primarily heterozygous and not associated
   with PTEN mutation. The significant association of the codeletions
   (i.e., PTEN and TP53) in PTCL-NOS with poor prognosis was noted in
   another study ([99]13). Although the canonical function of PTEN is to
   regulate PI3K activation through its phosphatase function,
   phosphatase-independent function was demonstrated in mice harboring a
   mutation causing loss of the PTEN C-terminal region ([100]14),
   resulting in a T cell acute lymphoblastic leukemia (T-ALL)–like
   disease. Similar PTEN C-terminal region mutations in patients with
   T-ALL were identified, resulting in truncation of the C2 domain without
   disrupting phosphatase activity ([101]15). PTEN promotes genomic
   stability via its direct interaction with TP53 ([102]16), and TP53 also
   directly regulates PTEN transcription ([103]17), thus having
   interdependent functions. Whereas mice with Pten deletions ([104]18)
   and Trp53 deletions ([105]19, [106]20) spontaneously develop T-ALL,
   mice with Trp53 deletion develop cytotoxic TCL with a CD1d-restricted
   natural killer (NK)/T cell of origin ([107]21) and a transcriptomic
   program similar to human hepatosplenic T cell lymphoma (HSTCL) and T
   cell prolymphocytic leukemia (T-PLL). The codeletion of TP53 and PTEN
   is an infrequent event in cancers but has been observed in advanced
   breast adenocarcinomas ([108]22), high-grade glioblastomas ([109]23),
   or aggressive prostate cancers ([110]24), and a murine model with
   codeletion demonstrated Myc-directed up-regulation of Phlpp2 to
   attenuate PI3K activity ([111]25).

   The co-occurring PTEN and TP53 functional loss (i.e., monoallelic PTEN
   loss with biallelic TP53 deletion/mutation) in PTCL-GATA3 is
   intriguing, but it is unclear how these two genetic events cooperate in
   T cell lymphoma pathogenesis, and the lack of authentic T cell or
   animal models to study PTCL-GATA3 pathogenesis is a major hurdle.
   Although the role of PI3K-AKT in thymic T cell selection, T cell
   activation ([112]26), effector T[H] cell function, polarization
   ([113]27), regulatory T cell (T[reg]) homeostasis, metabolic balance,
   and mitochondrial fitness ([114]28) has been explored, the functional
   relevance of PTEN in postthymic T cell neoplasms is unclear. TP53
   regulates antigen-specific responses via interleukin-2R (IL-2R)
   up-regulation ([115]29), homeostasis by up-regulating pro-apoptotic
   SLAM-associated protein (SAP) ([116]30), cytolytic activity, stemness,
   and T[reg] differentiation ([117]31), but the understanding of its role
   in T[H] cell biology is limited. Here, we investigated the role of
   these alterations independently and cooperatively in mature T cell
   differentiation and lymphomagenesis by evaluating PTCL-NOS biospecimens
   and establishing mouse models with Trp53^R172H mutation or Trp53
   deletion concurrent with Pten deletion in CD4+ T cells. Three models
   with Trp53 and Pten coaberration (Trp53^R172H;Pten^fl/+−,
   Trp53^R172H;Pten^fl/fl, and Trp53^fl/fl;Pten^fl/+) were highlighted,
   with the latter closely matching what is more frequently observed in
   PTCL-GATA3 cases. We also engineered normal human CD4+ T cells using a
   CRISPR-Cas9 methodology to delete TP53 or knock-in mutant TP53^R175H
   with PTEN deletion to investigate the human relevance of these genetic
   alterations. To enhance the translational relevance of these
   preclinical models, we compared the molecular features of mature T cell
   lymphomas (m-TCLs) with human PTCL (h-PTCL)–GATA3 and further
   investigated therapeutic interventions targeting codeleted m-TCLs.

RESULTS

TP53 and PTEN codeletion is associated with the PTCL-GATA3 subtype and poor
clinical outcome

   Earlier genomic copy number (CN) analysis (gCNA) studies showed that
   frequent codeletion of TP53 (del17p13) and PTEN (del10p23) was
   associated with PTCL-GATA3 (26%) and a subset of ALK-ALCL (15%) but not
   in other PTCL entities ([118]12, [119]32). TP53 and PTEN deletions were
   confirmed by fluorescence in situ hybridization (FISH) in a subset of
   PTCL-NOS (n = 33) ([120]Fig. 1A). Combined CN loss (TP53 and PTEN) was
   associated with inferior OS (P < 0.05) in PTCL-GATA3 treated with
   chemotherapy (CHOP/CHOEP) ([121]Fig. 1B). Among the 32 PTCL-GATA3 cases
   analyzed in our previous study ([122]12), loss of TP53 was observed in
   19 cases (59%) and PTEN loss in 11 cases (34%). Notably, nine of those
   cases (28%) exhibited codeletion of both tumor suppressors, indicating
   a substantial subset with dual gene loss (table S1). In contrast, among
   30 PTCL-TBX21 cases, only two and one case showed TP53 or PTEN loss,
   respectively (6.7 and 3.3%), with no codeletions identified. This
   marked difference underscores the genomic divergence between PTCL-GATA3
   and PTCL-TBX21, highlighting a potential molecular distinction between
   these subtypes. TP53 mutation co-occurred with del17p (5/20; 25%),
   mainly affecting the DNA binding domain, with the TP53^R175H variant
   being recurrent, whereas no PTEN mutations (0/20) were noted in
   PTCL-GATA3 ([123]Fig. 1C). PTEN and TP53 losses were significantly
   associated with down-regulated expression of their mRNAs ([124]Fig.
   1D).

Fig. 1. Evaluation of TP53 and PTEN codeletion in distinct PTCL-NOS subtype
(i.e., PTCL-GATA3).

   [125]Fig. 1.
   [126]Open in a new tab

   (A) Confirmation of TP53 and PTEN aberration by FISH in representative
   PTCL-NOS cases (of n = 33 where FISH was performed). The results
   confirmed previous observations ([127]12, [128]32). (B) OS analysis of
   PTCL-GATA3 cases with available outcome data by PTEN and TP53 DNA CN
   (CNA) status, suggesting poor outcome associated with double loss. Two
   cases with PTEN only loss were excluded from the plot due to a low
   number. (C) Frequency plot of DNA CN gains and losses in PTCL-GATA3
   affecting chromosomes 10 and 17 (top). Schematic of mutations
   identified affecting TP53 in PTCL-GATA3 (bottom). aa, amino acid. (D)
   mRNA expression of TP53 and PTEN genes in PTCL-GATA3 cases and
   association with gCNA of TP53 and PTEN showed significantly reduced
   mRNA expression with gCNA loss. (E) DEGs (P < 0.05; fold change > 2)
   between WT, del-TP53, del-PTEN, and TP53/PTEN codeleted in PTCL-GATA3
   cases. Representative transcripts are noted. The median expression for
   four comparisons is shown on the right. TP53/PTEN codeleted cases
   showed significant enrichment of a proliferative and PI3K/AKT
   transcriptional program. (F) Bubble plot of GSEA data comparing the
   del-TP53, del-PTEN, and codeleted TP53/PTEN versus WT [false discovery
   rate (FDR) < 0.025] showing significant enrichment of MYC and STAT3
   targets in codeleted cases. (G) Hematoxylin and eosin (H&E) and IHC
   staining for the noted biomarkers (i.e., GATA3, MYC, and Ki-67) in a
   representative case for TP53 and/or PTEN abnormality PTCL-GATA3 cases.
   (H) Validation of oncogenic MYC activation using IHC showed high MYC
   positivity in double loss cases, and GATA3 positivity was higher in
   TP53 loss cases. Statistical significance was determined using the
   Wilcoxon rank sum test. (I) Bubble plot of T cell subset proportions
   estimated by xCell ([129]35) in PTCL-GATA3, suggesting enrichment of
   T[H]2 immune cells in double deleted cases. NES, normalized enrichment
   score.

   Gene expression profiling (GEP) analysis of PTCL-GATA3 tumors with
   TP53/PTEN coaberrations (deletion/mutation) showed significant
   enrichment of a proliferative transcriptional program (i.e.,
   transcriptional targets of MYC and E2F), transcripts regulating G[2]-M
   checkpoint, or PI3K-mTORC1 signaling ([130]Fig. 1, E and F). Single
   del17p (TP53) tumors showed similar GEP with lower enrichment,
   including transcriptional signatures associated with T[H] cell
   differentiation (IL-4–induced targets and STAT3 targets) ([131]33). In
   contrast, wild-type (WT) cases were significantly enriched in
   SRC-induced signatures or KRAS-dependent/addicted tumor survival
   signatures ([132]34). PTCL-GATA3 tumors with coaberrant (TP53 and PTEN)
   had higher MYC mRNA and MYC-target gene signatures, which were
   validated by immunohistochemistry (IHC) showing higher MYC and Ki-67
   protein levels, the single TP53-del and WT cases had moderate and low
   MYC expression, respectively ([133]Fig. 1, G and H). PTCL-GATA3 cases
   with a TP53 loss showed higher GATA3 expression by IHC (P < 0.07),
   compared to TP53 WT cases. In general, MYC expression is higher in
   PTCL-GATA3 versus PTCL-TBX21 due to frequent 8q gain as shown
   previously ([134]12)[,] but codeleted tumors had even higher MYC
   expression. GEP deconvolution of immune-cell signatures using xCell
   ([135]35) indicated significant enrichment of T[H]2 cells and CD4+ T
   memory cells in tumors ([136]Fig. 1I). Overall, multiomics
   characterization of primary PTCL-GATA3 specimens indicated that MYC
   activation, T[H]2 differentiation, and inferior OS are associated with
   TP53/PTEN codeleted tumors.

Trp53^R172 knock-in or Trp53 knockout and Pten knockout in murine CD4+ T
cells lead to T cell malignancies

   Seven murine models, including WT (Cre−), Pten-heterozygous
   (Pten^Δ:Cd4-Cre^+/−;Pten^fl/+), Pten-homozygous
   (Pten^ΔΔ:Cd4-Cre^+/−;Pten^fl/fl), Trp53-mutant
   (Trp53^mut:Cd4-Cre^+/−;Trp53^LSL-R172H+/−), Trp53-mutant;
   Pten-heterozygous (Trp53^mutPten^Δ: Cd4-Cre^+/−;
   Trp53^LSL-R172H+/−;Pten^fl/+), and Trp53-mutant; Pten-homozygous
   (Trp53^mutPten^ΔΔ: Cd4-Cre^+/−;Trp53^LSL-R172H+/−; Pten^fl/fl)
   andTrp53-knockout; Pten-heterozygous (Trp53^ΔΔPten^Δ: Cd4-Cre^+/−
   Trp53^fl/fl;Pten^fl/+), were generated ([137]Fig. 2A). The Trp53^R172H
   mutation and Pten deletion were confirmed by polymerase chain reaction
   (PCR) of genomic DNA or from RNA sequencing (RNA-seq) and whole-exome
   sequencing (WES) of corresponding CD4+ T cells or murine tumors
   ([138]Fig. 2B). Trp53 deletion was confirmed by Transnetyx automated
   genotyping. The early onset of spontaneous lymphomas in double-mutant
   mice indicated that Pten loss accelerates tumor development, with a
   median OS of 10.5 weeks in Trp53^mutPten^ΔΔ (n = 9), 13.2 weeks in
   Pten^ΔΔ (n = 24), 22.4 weeks in Trp53^ΔΔPten^Δ (n = 7), 27.7 weeks in
   Trp53^mutPten^Δ (n = 33), and 49.7 weeks in Trp53^mut (n = 26), with
   several pathological manifestations including splenomegaly, enlarged
   thymus, or extranodal involvement (primarily in the liver) ([139]Fig.
   2, C and D). Whereas Pten^ΔΔ showed complete lymphoma penetrance, no
   tumor was observed in Pten^Δ after 2 years of follow-up, in agreement
   with earlier findings ([140]36), but Trp53^mut provided accelerated and
   cooperative oncogenic stimulation in Pten^ΔΔ or Pten^Δ mice (P < 0.05).
   Trp53^ΔΔPten^Δ mice developed aggressive tumors affecting the spleen
   and liver (fig. S1A) and developed tumor masses in the subcutaneous
   tissue, and several succumbed to a fatal illness other than lymphoma.

Fig. 2. Generation of conditional murine T cell lymphoma models.

   [141]Fig. 2.
   [142]Open in a new tab

   (A) Schematic of the characterization of different CD4+ T cell murine
   models and cross-breeding strategies of Cd4-Cre transgenic mice with
   Trp53^LSL-R172H+/−, Trp53^FL/FL, Pten^FL/FL designated as Pten^Δ,
   Pten^ΔΔ, Trp53^mutPten^ΔΔ, Trp53^mutPten^Δ, Trp53^mut, Trp53^mutPten^Δ,
   and Trp53^ΔΔPten^Δ. (B) Validation of Trp53 and Pten gene rearrangement
   in CD4+ T cells in murine models using standard PCR and SNV analysis in
   RNA-seq data. SNV, single nucleotide variant. (C) Necropsy of
   representative tumor-bearing mice with early onset in Pten^ΔΔ,
   Trp53^mutPten^ΔΔ, Trp53^ΔΔPten^Δ and Trp53^mut and late onset in
   Trp53^mut and Trp53^mutPten^Δ demonstrated splenomegaly and enlarged
   thymus, estimated by the corresponding weight analysis (bottom). (D) OS
   analysis of murine models. Kaplan-Meier curve analysis indicates that
   tumors with double loss have aggressive lymphoma and are associated
   with poor outcomes (Trp53^mutPten^ΔΔ, Trp53^ΔΔPten^Δ, and Pten^ΔΔ)
   compared to Trp53^mutPten^ΔΔ and Trp53^mut. (E) Morphologic and IHC
   characteristics in tumor-bearing mice. (F) Quantification of IHC
   staining of tumors or age-matched controls for CD3, CD4, CD8, and CD19,
   demonstrating a higher proportion of CD3, CD4, and CD8 splenocytes in
   tumor mice across most genotypes. (G) Clonality assessment using TCR
   repertoire analysis by RNA-seq in tumor-bearing mice compared to WT and
   preneoplastic mice. 4 wk mice, 4-week-old mice. (H) In silico analysis
   of subcellular fractions (using xCell) of the RNA-seq data indicated
   enrichment of the T[H]2 cellular profile in murine tumors affected by
   Trp53 or Pten aberration compared to WT mice. (I) Flow cytometry
   analysis and validation of T[H]2 immunophenotype of the murine tumors
   (gated on CD4+ T cells) indicated high GATA3 expression in the
   Trp53^ΔΔPten^Δ, Trp53^mutPten^Δ, and Pten^ΔΔ tumors compared to
   age-matched WT CD4+ T cells. The flow plots are one representative
   mouse, and the bar graph on the right shows the GATA3+/TBX21−
   percentage for three mice per genotype profiled.

   Whole-tissue sections or tissue microarrays (TMAs) with ~3 to 5
   representative tumors for each genotype were immunoassayed for
   immunological and maturation biomarkers (i.e., CD3, CD4, CD8A, and
   TDT), and the pathological findings and immunophenotypic data are
   summarized in table S2. Tumors involved the thymus, spleen, and liver
   and had an atypical pattern of highly proliferative, large CD3^+ T
   cells that were single positive (SP; CD4^+ or CD8^+) or double positive
   (DP; CD4^+CD8^+) and TDT negative, indicating m-TCLs. Whereas CD4+ T
   cell neoplasms were dominant in Trp53^mutPten^ΔΔ and Pten^ΔΔ mice and
   had similar histological and immunophenotypic features as reported
   earlier ([143]37), Trp53^mutPten^Δ mice also developed CD8^+ T cell
   lymphomas ([144]Fig. 2, E and F). Trp53^ΔΔPten^Δ mice developed m-TCLs
   that were TDT negative, with four of seven mice indicating DP
   (CD4^+CD8^+) T cells and three of seven SP (CD4^+) T cells upon IHC
   examination.

   The clonal assessment with T cell receptor (TCR) repertoire analysis
   using RNA-seq or flow cytometry identified a major TCR-α or TCR-β clone
   in murine-TCL but not in age-matched WT CD4+ T cells, suggesting a
   restricted TCR repertoire and clonal progression ([145]Fig. 2G and fig.
   S2, A and B). In silico analysis of RNA-seq expression profiling of
   neoplastic cells suggested T[H]2-enriched gene expression in
   Trp53^mutPten^Δ and Trp53^ΔΔPten^Δ ([146]Fig. 2H). Flow cytometry
   analysis validated higher GATA3^+ CD4^+ T cell in Trp53^mutPten^Δ
   (29%), Trp53^ΔΔPten^Δ (70%), and Pten^ΔΔ (23%) versus age-matched WT
   cells (6.7%) ([147]Fig. 2I).

   Splenomegaly in tumor-bearing mice was associated with varied
   splenocyte subset cellular distribution compared to WT. Some
   Trp53^mutPten^Δ mice had decreased CD4+ T cells and an increase in CD8+
   T cells compared to age-matched WT controls (fig. S1A), whereas
   corresponding IHC indicated infiltration of B cells in Trp53^mutPten^Δ
   ([148]Fig. 2E). Enlarged thymuses were also noted, and thymic cellular
   characterization demonstrated a significant increase in SP (CD4+) and
   reduced DP (CD4+CD8+) T cells in tumors with Pten^ΔΔ (i.e.,
   Trp53^mutPten^ΔΔ or Pten^ΔΔ). A subset (two of seven) of Trp53^mut mice
   had reduced DP and increased double negative (DN) and SP (CD4 or CD8
   markers) (fig. S1A), which may be due to depletion or negative
   selection of CD4+ thymocytes, as demonstrated in an earlier study
   ([149]38). The evaluation of the cellular proportions before
   lymphomagenesis at 6- and 10-week timepoints revealed no significant
   changes in either B cells, NK cells, myeloid cells, CD3^+ T cells, or
   CD4+/CD8+ T cell subpopulations in splenocytes (fig. S1B).

Molecular characterization of murine lymphomas indicates a T[H]2-like
transcriptomic program

   We performed RNA-seq analysis of CD4+ T cells isolated from 4-week-old
   mice before lymphoma development (termed as preneoplastic), neoplastic
   splenocytes, and age-matched WT cells from the above murine models
   ([150]Fig. 3A). Unsupervised principal components analysis (PCA) and
   hierarchical consensus clustering (HCC) indicated that neoplastic and
   preneoplastic CD4+ T cells cluster separately (fig. S2, C and D). GEP
   analysis of the CD4+ T cells with Pten^ΔΔ, Trp53^mutPten^ΔΔ or
   Trp53^ΔΔPten^Δ showed a higher number of differentially expressed
   transcripts compared to either Trp53^mutPten^Δ or Trp53^mut versus
   their corresponding age-matched WT CD4+ T cells, suggesting that
   complete loss of p53- or Pten-mediated signaling has a substantial
   impact on global transcriptional activity. Similar observations were
   noted in early-occurring m-TCLs compared to late-occurring m-TCLs (fig.
   S2E). In silico functional analysis [i.e., using the Kyoto Encyclopedia
   of Genes and Genomes (KEGG) database] of the differentially expressed
   transcripts indicated distinctive and shared transcriptomic signatures
   associated with these aberrations in preneoplastic CD4^+ T cells
   including up-regulation of transcripts promoting cell cycle, TGF-β
   (transforming growth factor–β), or RAP1 (endocytosis) pathways and
   down-regulation of transcripts regulating extracellular matrix
   reorganization and cell adhesion ([151]Fig. 3B). CD4^+ T cells with
   Pten^ΔΔ, Trp53^mutPten^ΔΔ, Trp53^mutPten^Δ, and Trp53^ΔΔPten^Δ showed
   significant differential gene expression (DGE) of transcripts
   regulating pathways promoting T cell activation and differentiation
   [i.e., ROS (reactive oxygen species), TCR, JAK-STAT (Janus
   kinase–signal transducer and activator of transcription), NF-κB
   (nuclear factor κB), and T[H] cell differentiation] and WNT signaling
   in Trp53^ΔΔPten^Δ (see the gene list in [152]Fig. 3B). Gene Set
   Enrichment Analysis (GSEA) validated and demonstrated significant
   enrichment of RNA signatures associated with T cell activation, T[H]
   differentiation, STAT3, PI3K-responsive, and MYC-regulated
   transcriptome in Trp53^mutPten^ΔΔ, Trp53^mutPten^Δ, Pten^ΔΔ and
   Trp53^ΔΔPten^Δ ([153]Fig. 3C). We validated representative pathways
   including PI3K signaling, i.e., p-AKT (Ser^473), STAT3 activation
   (p-STAT3–Y705), and T cell activation [p-LCK (Y505) and p-ZAP70
   (Y319)], at the protein level by using Western blot in another mouse
   cohort ([154]Fig. 3D).

Fig. 3. RNA-seq analysis in neoplastic CD4+ T cells at the preneoplastic
stage and neoplastic stage and validation at protein levels using either
Western blot or flow cytometry in an independent murine cohort.

   [155]Fig. 3.
   [156]Open in a new tab

   (A) Schematic of the experimental design for functional
   characterization of preneoplastic CD4+ T cells and neoplastic cells
   isolated from mouse models. (B) RNA-seq analysis of the CD4+ T cells
   isolated from mouse models and heatmap depicting the DEGs versus WT
   before development of lymphoma (4 weeks). (C) Bubble plot of GSEA
   indicated perturbed pathways (T cell activation, T[H] differentiation,
   STAT3, PI3K-responsive, and MYC-regulated transcriptome) in
   preneoplastic CD4+ T cells compared to their WT counterpart. (D)
   Protein expression validation of key oncogenic genes including TCR,
   PI3K-AKT signaling, and STAT3 signaling by Western blot in an
   independent mouse cohort, indicating increases in TCR activation, STAT3
   and PI3K/AKT signaling in mutant mice compared to WT. (E) Heatmap of
   the DEGs in tumors from the noted mouse models versus age-matched
   spleens. The preneoplastic cells are shown for comparison. Enriched
   pathways are indicated in the bubble plot on the left. (F) Bubble plot
   of select GSEA pathway enrichment analysis results of the neoplastic
   spleen compared to their WT age-matched counterpart. (G) IHC validation
   of p-STAT3 protein in tumor-bearing mice versus their corresponding WT
   controls. IHC quantification was performed on all biological replicates
   (n = 3) and shown as a bar chart.

   RNA-seq analysis of the m-TCLs versus their corresponding age-matched
   splenocytes indicated up-regulation of proliferation-promoting
   transcriptional signatures [e.g., cell cycle transition, DNA
   replication, RNA metabolism, and mammalian target of rapamycin (mTOR)
   pathway] and transcripts regulating metabolic
   reprogramming/thermogenesis, DNA damage, and aerobic mitochondrial
   metabolism, consistent with earlier findings ([157]39) ([158]Fig. 3, E
   and F). We evaluated transcriptome associated with oncogenic transition
   by performing DGE analysis between m-TCLs and their corresponding
   preneoplastic CD4+ T cells collected at an earlier timepoint (i.e., 4
   weeks) and age-matched WT splenocytes. Whereas m-TCLs with Pten loss
   (i.e., Pten^ΔΔ, Trp53^mutPten^ΔΔ, Trp53^mutPten^Δ, and Trp53^ΔΔPten^Δ)
   showed the concordant genes/pathway promoting cell cycle, mitosis, and
   DNA repair, the m-TCLs from Trp53^mut showed dysregulations in the
   chemokine signaling (fig. S3). Specifically, transcripts regulating
   TCR, STAT3, and T[H]2 associated (GATA3) signaling were enriched in
   Trp53^mutPten^ΔΔ, Pten^ΔΔ, Trp53^mutPten^Δ, and Trp53^ΔΔPten^Δ m-TCLs
   and with down-regulation of IL-12 signaling, a major regulator of
   TBX21/T[H]1 signaling ([159]Fig. 3, E and F). Validation of the GATA3
   protein expression was performed by flow cytometry ([160]Fig. 2I) and
   p-STAT3 by IHC in additional m-TCLs ([161]Fig. 3G). These findings in
   double-mutant m-TCLs show a T[H]2-like transcriptional program and
   immunophenotype, and STAT3 activation may induce GATA3 expression, as
   shown earlier ([162]33).

   We performed WES on several m-TCLs to uncover additional genetic
   drivers and revealed that four mice displayed gains or amplifications
   involving the Myc oncogene, in particular with Pten^ΔΔ. In addition,
   mutation of Stat2 (Pten^ΔΔ); Mybbp1a, DNA repair regulation
   (Trp53^mutPten^Δ); Bcl2l10, anti-apoptotic regulator (Trp53^mut), Dlx5,
   a Myc regulator (Trp53^mutPten^ΔΔ) and Pi3kr2, and PI3K modulator
   (Trp53^ΔΔPten^Δ) were identified, suggesting additional genes involved
   in Trp53/Pten-driven lymphomagenesis (fig. S2F).

Trp53 and Pten abnormalities drive CD4+ T cell activation and T[H]2
differentiation ex vivo

   Proliferation, cell cycle, apoptosis, and plasticity of the CD4+ T
   cells isolated from splenocytes of each mouse genotype at 4 weeks were
   compared with those of WT cells under normal in vitro culture
   conditions (i.e., with/without α-CD28/CD3 and IL-2) ([163]Fig. 4A).
   Trp53^mutPten^Δ had a growth advantage compared to WT under normal
   conditions and reduced serum (5%) conditions, whereas the cells with
   other genotypes, including Trp53^mutPten^ΔΔ, showed reduced growth
   compared to WT ([164]Fig. 4, B and C). Under low (10 ng) or normal (20
   ng) IL-2–only conditions, however, both Trp53^mutPten^Δ,
   Trp53^ΔΔPten^Δ, and Trp53^mutPten^ΔΔ CD4+ T cells survived and expanded
   without CD28/CD3 stimuli. Trp53^mutPten^ΔΔ cells demonstrated their
   growth responsiveness to elevated IL-2 levels, highlighting their
   reliance on IL-2, but not CD3+CD28 stimulation for proliferation.
   Pten^ΔΔ and Trp53^mutPten^ΔΔ proliferated the slowest and had increased
   apoptosis compared to other genotypes and WT ([165]Fig. 4C).

Fig. 4. In vitro characterization of CD4+ T cells with Trp53 or Pten genetic
lesions.

   [166]Fig. 4.
   [167]Open in a new tab

   (A) Schematic of the experimental designs for functional
   characterization of the CD4+ T cells with Trp53 or Pten loss. CD4+ T
   cells were isolated by negative selection from 4-week-old mice and
   cultured under various conditions to assess growth and survival
   characteristics. Four-week-old T cells were also transplanted into WT
   mice to assess their fitness in vivo. FBS, fetal bovine serum. (B)
   Proliferation of CD4+ T cells from the different mouse models under
   differing culture conditions, where Trp53^mutPten^Δ had a proliferative
   advantage under serum-starved conditions, and double-mutant cells had
   an advantage under IL-2–only conditions. (C) Apoptosis (annexin V and
   7-AAD) was analyzed by flow cytometry. Pten loss was found to be
   associated with a higher apoptotic rate in these murine preneoplastic
   CD4+ T cells. 7-AAD, 7-aminoactinomycin D. (D) The steady-state
   engraftment of CD45.2+ cells was routinely detected in PBMCs obtained
   through maxillary bleeding 2 weeks postadoptive transfer. The right
   panel shows the flow cytometry profile of the CD45.2+ cells. (E)
   Schematic representation of experiments assessing T[H]1 or T[H]2
   polarization in preneoplastic CD4+ T cells. (F) Representative Western
   blot of GATA3 and TBX21 in preneoplastic CD4+ T cells cultured under
   T[H]1 or T[H]2 conditions versus naïve conditions. Western blot
   quantification was performed on three biological replicates and shown
   as the bar graph on the right. 0 hr, 0 hours. (G) Representative flow
   cytometry analysis of Gata3 and Tbx21 in Trp53^mutPten^Δ,
   Trp53^ΔΔPten^Δ, and age-matched WT CD4+ T cells cultured in T[H]1- and
   T[H]2-polarizing media, suggesting a T[H]2-promoting immunophenotype in
   Trp53^mutPten^Δ. The bar graph depicts three biological replicates and
   shows GATA3+/TBX21− and GATA3−/TBX21+ for T[H]2 and T[H]1 conditions,
   respectively, compared to 0 hours. (H) Proliferation analysis of the
   preneoplastic CD4+ T cells under the T[H]1 or T[H]2 conditions
   indicates that Trp53^mutPten^Δ showed a survival advantage under both
   conditions compared to WT cells.

   To assess fitness under in vivo conditions, we performed adoptive
   transfer studies using CD4^+ T cells isolated from 4- to 5-week-old
   (preneoplastic) mice into 10-week-old recipient mice (postthymic
   development). Pten^ΔΔ or Trp53^mutPten^ΔΔ CD4+ T cells were able to
   engraft, survive, and proliferate for at least 12 weeks and outcompete
   recipient CD4+ T cells, whereas Trp53^mut and Trp53^mutPten^Δ donor
   CD4+ T cells spontaneously regressed in host mice ([168]Fig. 4D).
   Tumors occurred in mice engrafted with Pten^ΔΔ or Trp53^mutPten^ΔΔ CD4+
   T cells at 3.4 months postengraftment, indicating that PI3K activation
   is required for engraftment fitness and is indicative of clonal
   expansion. The engrafted Pten^ΔΔ tumor-bearing mice show CD4+CD8+ T
   cell subpopulation in the thymus, spleen, and blood, whereas
   Trp53^mutPten^ΔΔ tumor-bearing mice showed a higher percentage of DP
   cells in the thymus and blood (fig. S1C).

   To assess the role of Trp53^mut and/or Pten deletion in T[H] cell
   differentiation, naïve CD4+ T cells were isolated from each murine
   genotype and stimulated under exogenous T[H]1- and T[H]2-polarizing
   conditions ([169]40) ([170]Fig. 4E). GATA3 protein expression increased
   substantially in Trp53^mut, Trp53^mutPten^Δ, and Trp53^mutPten^ΔΔ,
   indicating preferential polarization toward the T[H]2 phenotype
   compared to corresponding WT cells, a finding consistent with
   corresponding transcriptomic analysis ([171]Fig. 4, F and G).
   Furthermore, flow cytometry analysis of the double-mutant cells
   suggested that Trp53^ΔΔPten^Δ cells had higher basal GATA3 protein
   expression compared to corresponding WT cells and were resistant to
   T[H]1 polarization ([172]Fig. 4G). The Trp53^mutPten^Δ cells showed
   better growth in T[H]2 medium compared to other mouse model genotypes,
   especially in long-term culture (P < 0.05) (day 10) ([173]Fig. 4H). The
   observation that CD4+ T cells from Trp53^mut, Trp53^mutPten^Δ,
   Trp53^mutPten^ΔΔ, and Pten^ΔΔ had elevated GATA3 expression compared to
   WT when grown under T[H]2 conditions and reduced TBX21 expression
   compared to WT when grown under T[H]1 conditions ([174]Fig. 4F) is
   quite intriguing as CD4+ T cells from C57BL/6 mice preferentially
   produce T[H]1 cytokines with high interferon-γ (IFN-γ) and low IL-4
   ([175]41). Thus, Trp53/Pten may regulate T[H] lineage plasticity and
   differentiation.

CRISPR-edited human CD4+ T cells with aberrant TP53 and PTEN signaling
favored T[H]2 differentiation

   We engineered human CD4+ T cells from four healthy donors to generate
   genotypes similar to the murine models, including TP53^ΔΔ, TP53^R175H,
   PTEN^ΔΔ, PTEN^Δ, and double-mutant (TP53^ΔΔPTEN^Δ and TP53^ΔΔPTEN^ΔΔ)
   using a CRISPR-Cas9 approach ([176]42) ([177]Fig. 5A and fig. S4A).
   TP53^ΔΔ CD4+ T cells exhibited multinucleated cells (fig. S4B). Loss of
   TP53 and down-regulation of target genes (e.g., CDKN1A/P21 and BAX) and
   knockout of PTEN and PI3K activation (i.e., p-AKT) due to PTEN loss
   were validated at the protein or mRNA level ([178]Fig. 5B and fig.
   S4C). CD4+ T cells with TP53^ΔΔ exhibited the ability to proliferate
   both under normal conditions (CD3/CD28+ IL-2) and in the absence of
   IL-2 (fig. S4D), but cells with PTEN^ΔΔ were difficult to maintain in
   culture and underwent rapid apoptosis upon TCR activation (fig. S4E),
   consistent with findings in murine CD4+ T cells, indicating that PI3K
   hyperactivation is disadvantageous. CD4+ T cells with TP53^ΔΔPTEN^Δ
   displayed the capability to grow in IL-2 stimulation alone, without
   dependence on CD3 and CD28 activation. Notably, when cultured in IL-2
   with CD3 and CD28 stimulation, these cells did not proliferate well,
   whereas PI3K inhibition by duvelisib treatment improved their survival,
   indicating that regulation of PI3K activation is important for T cell
   survival (fig. S4D).

Fig. 5. Generation and molecular analysis of human-edited CD4+ T cells.

   [179]Fig. 5.
   [180]Open in a new tab

   (A) Schematic representation showing an experimental approach for the
   CRISPR-Cas9 engineered human CD4+ T cells. GAPDH,
   glyceraldehyde-3-phosphate dehydrogenase. (B) Validation of TP53 and/or
   PTEN knockout (KO) models. Reduced expression of PTEN and TP53 was
   confirmed by Western blot, and increased phosphorylation of PTEN target
   AKT was confirmed in PTEN^ΔΔ or PTEN^Δ CD4+ T cells. (C) Heatmap of the
   DEGs in human CD4+ T cell (modified cells). (D) GSEA of the
   CRISPR-modified cells versus their Cas9 controls. (E) Representative
   (of three) flow cytometry analyses for GATA3 and TBX21 expression in
   TP53^ΔΔ and TP53^R175H cells showed higher GATA3+ cells versus Cas9 WT
   control. (F) Validation of the TCR activation by Western blotting of
   TP53^ΔΔ CRISPR-modified cells. (G) TP53 ChIP-seq data (GSE #86164) show
   the enrichment of the region within the GATA3 gene (intron-3) in two
   cell lines upon treatment with Nutlin-3 (top). (H) ChIP-qPCR of the
   indicated regions in the GATA3 promoter and intron region show
   significant enrichment in the intron-3 region. The promoter region of
   CDKN1A (p21) was used as a control. Negative control, intron-3, and p21
   fold enrichment was also determined in SMZ1, TP53^R175H CD4+ T cell
   knock-in, and TP53^∆∆ CD4+ T cell knockout, which showed inhibited or
   reduced enrichment compared to 293T (WT TP53). (I) Representative flow
   cytometry analysis of ectopic TP53 expression in 293T cells resulting
   in reduced GATA3 expression. GFP, green fluorescent protein. (J)
   Firefly and Renilla luciferase reporter assay with cotransfection of
   TP53 expression vectors showed repression of luciferase activity
   compared to the control vector. EV, empty vector; F.Luc, firefly
   luciferase; R.Luc, Renilla luciferase; RPM, reads per million/kb.

   We performed RNA-seq analysis of the edited human CD4+ T cells versus
   their corresponding control Cas9 cells. Unsupervised hierarchical
   clustering indicated that edited T cells primarily clustered by donor
   and culture time (fig. S4F), and TCR analysis showed restricted TCR
   clonality in the modified cells, which tended to show greater
   inclination toward oligoclonal evolution over time in culture (fig.
   S4G). We compared gene signature enrichment from the CRISPR-modified
   CD4+ T cells to their murine CD4+ T cell counterparts. Gene signatures
   enriched in TP53^ΔΔ and TP53^R175H CD4+ T cells showed similarities to
   each other and to their mouse counterparts, including T[H]2 enrichment
   and STAT activation ([181]Fig. 5, C and D). TP53^ΔΔ/PTEN^Δ human and
   mouse cells had enrichment of T[H]2, AKT, STAT, and IL-2
   response-related signatures, whereas PTEN^Δ human and mouse cells had
   enrichment of T[H]2 and PI3K-responsive signatures. Human CD4+ T cells
   with TP53^ΔΔPTEN^ΔΔ and PTEN^ΔΔ were extremely difficult to expand in
   culture and were cultured in the absence of TCR engagement,
   demonstrating an overlap with their murine counterpart and had reduced
   expression of proliferation-related signatures such as MYC.

   In agreement with murine findings, and corroborating the GEP signature
   finding, a higher percentage of GATA3+ (%) T cells were observed in
   TP53^ΔΔ (66% versus 30%; P < 0.05) and TP53^R175H (48% versus 34%)
   compared to their Cas9 corresponding control cells ([182]Fig. 5E). We
   further evaluated the expression of the GATA3 target gene CCR4 by flow
   cytometry and observed higher expression of GATA3+/CCR4+ in TP53^ΔΔ
   (44%; P < 0.05) and TP53^R172H KI (33%) compared to Cas9 control cells
   (21%) (fig. S4H). We validated the GEP findings, [i.e., enhanced TCR
   activity (p-LCK (Py505) and p-ZAP70(pY319)], PI3K-p85(Y458), and mTORC1
   (Thr^37/46) in TP53^ΔΔ by using Western blot or by flow cytometry
   ([183]Fig. 5F and fig. S4, I and J). Although CD4+ T cells with TP53^ΔΔ
   showed a growth advantage in T[H]2 conditional medium, their growth was
   inhibited in T[H]1 medium (fig. S4K). PTEN^ΔΔ or PTEN^Δ cells did not
   survive under any conditional media supplement, illustrating that
   hyperactivation of the PI3K pathway due to PTEN loss may be detrimental
   for in vitro proliferation.

TP53 directly regulates GATA3 expression and thus T[H]2 differentiation

   Murine and human CD4+ T cells showed up-regulated GATA3 expression upon
   TP53 loss ([184]Figs. 4, F and G, and [185]5E). We examined TP53
   chromatin immunoprecipitation sequencing (ChIP-seq) datasets (see
   Materials and Methods; [186]Fig. 5G) and observed potential TP53
   binding regions in the GATA3 proximal (PP[1] and PP[2]:
   chr10:8,09,6687-7460) and distal promoter (DP; chr10:8,087,212-410) and
   intron-3 region (chr10:8102,082-325) [ChIP–quantitative PCR (qPCR)]
   ([187]Fig. 5G). In silico analysis of the TP53 binding site showed a
   single decamer TP53 binding site at the intron-3 locus of GATA3,
   lacking one nucleotide but functional in an earlier study ([188]43).
   The TP53 binding motif (RRRCWWGYYY) is present in GATA3 intron-3 using
   CentriMo Local Motif Enrichment Analysis ([189]43) but lacking the
   third “Y.” The motif investigated is the only TP53 motif with a MEME
   motif score of ≥5 in GATA3 promoter or intronic regions. To validate
   these findings further, TP53 was ectopically expressed in 293T and TP53
   ChIP-qPCR was performed to interrogate the GATA3 locus. ChIP-qPCR
   results showed high (>5-fold) enrichment of p53 binding at the promoter
   regions (PP[1] and PP[2]) and more than 10-fold enrichment at intron-3
   compared to isotype control ([190]Fig. 5H). Furthermore, TP53
   expression in the 293T cells resulted in lower GATA3 expression
   ([191]Fig. 5I). Further validation in the SMZ1 T cell line carrying
   TP53^F113S showed >10-fold enrichment in the intron-3 locus ([192]Fig.
   5H). This enrichment at intron-3 was not observed in CD4+ T cells with
   TP53^R175H or TP53^ΔΔ, which also have elevated GATA3 expression
   relative to CD4+ T cells with WT TP53 ([193]Fig. 5E). The positive
   control (CDKN1A/P21 promoter) did retain partial binding in TP53^R175H
   cells but not in TP53^ΔΔ ([194]Fig. 5H).

   We cloned three target regions [i.e., distal promoter, intron-3 region,
   and CDKN1A/P21 promoter region (positive TP53 binding control)] into a
   firefly luciferase reporter and cotransfected with TP53 expression
   vectors and SV40 Renilla luciferase reporter in 293T cells. The
   reporter assay showed repression of luciferase activity by TP53
   overexpression in the intron-3 region compared to empty vector,
   suggesting that TP53 may directly regulate GATA3 expression ([195]Fig.
   5J). In addition, a 50–base pair (bp) region of the GATA3 intron-3
   containing TP53 binding motif was cloned into firefly luciferase
   reporters, and repression of luciferase activity similar to that of the
   full intron-3 construct was observed. Upon mutating the motif, the
   luciferase activity was increased by eightfold, suggesting direct TP53
   binding and repression of GATA3 at this locus ([196]Fig. 5J).

The m-TCL transcriptomic signature is prognostic in h-PTCL

   We next identified the overlapping mouse and human transcripts to
   compare the hPTCL-GATA3 tumors to the m-TCLs to evaluate their
   association with each other ([197]Fig. 6A). We first assessed the
   expression of the human GATA3 classifier signature genes ([198]5) in
   the murine models. Whereas the PTCL-GATA3-(up) signature genes were
   either significantly enriched in Trp53^mutPten^Δ (P = 0.03) or trending
   Trp53^ΔΔPten^Δ (P = 0.07) m-TCL by GSEA, the PTCL-GATA3-(down) showed
   low expression in Trp53^mutPten^Δ, Trp53^ΔΔPten^Δ, Trp53^mutPten^ΔΔ,
   and Pten^ΔΔ ([199]Fig. 6B), suggesting a similarity between the m-TCL
   and hPTCL-GATA3, with a frequent codeletion of TP53 and PTEN.

Fig. 6. Integrative transcriptomic analysis of human (h-PTCLs) and murine
(m-TCLs).

   [200]Fig. 6.
   [201]Open in a new tab

   (A) Schematic representation of the ortholog analysis of h-PTCL and
   m-TCL transcriptomics. (B) Heatmap depicting the human GATA3
   classifier ([202]5) in the m-TCLs. TP53^mutPten^Δ tumors notably show
   enriched PTCL-GATA3-high signature genes (P = 0.03), resembling
   hPTCL-GATA3. (C) Consensus DEGs between mouse TP53^mutPten^Δ tumors and
   hPTCL-GATA3 cases with TP53 and PTEN loss. (D) Consensus DEGs between
   mouse TP53^ΔΔPten^Δ tumors and hPTCL-GATA3 cases with TP53 and PTEN
   loss. (E) Consensus DEGs between mouse TP53^mutPten^ΔΔ tumors and
   hPTCL-GATA3 cases with TP53 and PTEN loss were identified and OS in
   another PTCL-NOS cohort on the basis of the expression of the
   signature. (F) Consensus DEGs between mouse Pten^ΔΔ tumors and
   hPTCL-GATA3 cases with PTEN loss were identified. (G) Consensus DEGs
   between mouse Trp53^mut tumors and hPTCL-GATA3 cases with TP53 loss
   were identified. (H) The consensus DE signatures were evaluated in
   another PTCL-NOS cohort and cases were separated into halves on the
   basis of differential signature expression and evaluated for OS. The
   bar plot shows the PTCL-subtype composition of the cases separated by
   low and high signature expression, indicating that PTCL-GATA3 cases
   were enriched in high expression signature.

   Next, we identified the differentially expressed genes (DEGs) in each
   m-TCL or the preneoplastic CD4+ T cells from each genotype, compared to
   either corresponding WT controls, and PCA of these DEG signatures in
   hPTCL-NOS cases indicated that these signatures were able to separate
   the PTCL-GATA3 cases into a cluster distinct from PTCL-TBX21 cases
   (fig. S5, A to K). Furthermore, we generated consensus gene signatures
   between the m-TCL genotypes and PTCL-GATA3 cases carrying the
   corresponding TP53/PTEN deletion ([203]Fig. 6, C to G) and evaluated
   whether these refined transcriptomic signatures can demonstrate
   diagnostic or prognostic significance in independent PTCL-NOS cohort
   from earlier GEP studies ([204]5, [205]6, [206]44). We observed that
   the gene signatures were able to subclassify PTCL-NOS cases into
   PTCL-GATA3 and PTCL-TBX21 subtypes (P < 0.05; [207]Fig. 6H, left),
   suggesting that m-TCLs recapitulate molecular pathogenesis of the
   PTCL-GATA3 subtype. Furthermore, the signatures demonstrated prognostic
   significance in this independent cohort as cases with higher
   differential signature expression of the Trp53^mutPten^Δ,
   Trp53^ΔΔPten^Δ, and the Trp53^mutPten^ΔΔ consensus signatures trended
   toward worse outcomes ([208]Fig. 6H, right).

m-TCLs with combined TP53 and PTEN loss are therapeutically targetable

   We allografted Trp53^mutPten^ΔΔ and Trp53^ΔΔPten^Δ neoplastic
   splenocytes (P[0]) subcutaneously into nonobese diabetic severe
   combined immunodeficient gamma (NSG) mice to develop a cohort of tumors
   for preclinical assessment ([209]Fig. 7A). Tumor development in NSG
   mice was observed within a week at the site of inoculation, and tumors
   emerged in the secondary lymphoid organs around 10 days, requiring
   euthanasia by 30 days (P[1]). The P[1] tumor cells were adoptively
   transferred to generate P[2] tumors in NSG mice, and P[2] tumor
   formation was highly aggressive, with a 2-week survival time. We
   confirmed the Trp53 mutation and Pten deletion in secondary passages
   and confirmed the clonal relationship of the tumors from the P[0] and
   P[1], by Trp53 mutation and Pten deletion genotyping using Trp53/Pten
   rearrangement primers (table S3 and fig. S6A). In addition, IHC and
   analysis by flow cytometry showed consistent immunophenotype, i.e.,
   CD3+, CD4+, CD8−, p-STAT3, and TDT− ([210]Fig. 7B). We evaluated GATA3
   expression in successive generations ([211]Fig. 7C).

Fig. 7. Preclinical assessment of Pten/Trp53 loss-derived tumors with PI3K
inhibitors.

   [212]Fig. 7.
   [213]Open in a new tab

   (A) Schematic of Trp53^ΔΔPten^ΔΔ and Trp53^mutPten^ΔΔ tumors in NSG
   mice (P[1] and P[2]) using tumor splenocytes from parental tumor
   (P[0]). P1 tumors appeared in 4 to 6 weeks post-P[0] engraftment,
   whereas P[2] tumors appeared in 2 weeks post-P1 engraftment in NSG
   mice. (B) IHC analysis of the P[0], P[1], and P[2] Trp53^mutPten^ΔΔ and
   Trp53^ΔΔPten^Δ tumors reaffirm the expression of CD3, CD4, and CD8 in
   tumor cells and no TDT expression. (C) Representative flow cytometry
   analysis showed that GATA3 expression was elevated compared to WT-NSG
   (n = 3) in P[0] and P[1] Trp53^mutPten^ΔΔ (n = 6) and Trp53^ΔΔPten^Δ
   (n = 3) samples. (D) Kaplan-Meier plot displaying the OS advantage in
   P2 mice treated with duvelisib compared to vehicle control. (E)
   Duvelisib-treated mice had smaller observed spleens than vehicle
   control mice. (F) Kaplan-Meier plot displaying the OS advantage in P2
   mice treated with copanlisib compared to vehicle control. (G)
   Copanlisib-treated mice had smaller observed spleens than vehicle
   control mice.

   To test the effectiveness in vivo, P[2] mice generated from
   Trp53^mutPten^ΔΔ and Trp53^ΔΔPten^Δ were each randomized into two
   groups (n = 6 each) and treated with clinical-grade duvelisib (10
   mg/kg) and copanlisib (10 mg/kg) compared with vehicle treatment
   control. We observed a survival advantage with reduced splenomegaly in
   mice treated with either duvelisib (P < 0.001) ([214]Fig. 7, D and E)
   or copanlisib (P = 0.005) ([215]Fig. 7, F and G). Although the two
   inhibitors target different PI3K components, we analyzed a few spleens
   after treatment by IHC and observed decreased p-STAT3 expression (fig.
   S6, B and C). These data suggest that aberrant p53 signaling, a
   hard-to-target pathway, is vulnerable to PI3K inhibition in preclinical
   models.

DISCUSSION

   Recent multiomics studies have delineated two biological subtypes
   within PTCL-NOS, with the PTCL-GATA3 subtype frequently associated with
   codeletion of TP53 and PTEN and a poor outcome ([216]2, [217]5, [218]6,
   [219]44). GATA3, known for its role in T[H] cell differentiation, also
   regulates T[H]2 cell growth and proliferation and has been also shown
   to have a proto-oncogenic role in T cell neoplasms ([220]10, [221]45).
   We generated CD4+ T cell conditional murine models with hotspot
   mutation Trp53^R172H or Trp53 loss (Trp53^mut or Trp53^ΔΔ) combined
   with Pten loss (Pten^Δ) and observed CD4+ T cell m-TCLs ([222]37) with
   GATA3 immunophenotype and enriched with T[H]2-like GEP. Trp53^mutPten^Δ
   promoted lymphoma development with increased thymic or splenic
   cellularity and a significantly higher number of CD4+ T cells, B cells,
   and myeloid cells. The altered cellularity with a significant increase
   in B cells in the thymus may be attributed to depletion or negative
   selection of CD4+ thymocytes ([223]38), as noted in Pten-deficient
   HY-TCR transgenic mice ([224]46). Trp53^mutPten^ΔΔ showed accelerated
   lymphoma development with significant enrichment of CD4+ T cells in the
   lymphoid organs, consistent with previous findings ([225]47). Similar
   histological and molecular observations were noted in Trp53^ΔΔPten^Δ,
   thus reaffirming the cooperative role of the two genes in
   lymphomagenesis. Trp53^mutPten^Δ and Trp53^ΔΔPten^Δ mice also developed
   m-TCLs with including DP CD4+CD8+ immunophenotype. The DN (CD4^−CD8^−)
   to DP (CD4^+CD8^+) transition during T cell maturation is dependent on
   the pre-TCR expression and negatively regulated by Trp53 ([226]48);
   thus, enriched DP immunophenotype in some neoplasms is expected and
   reaffirming its role in the maturation phase.

   These neoplasms were further characterized with RNA-seq, indicating a
   T[H]2-like transcriptomic program, characterized by high Gata3 mRNA and
   protein expression and enrichment of its target genes. These neoplasms
   also demonstrated activated TCR, STAT3 signaling, and MYC-regulated
   transcriptome, where the loss of TP53 regulation could be driving this
   phenotype ([227]49, [228]50). TP53-directed STAT3-MYC oncogenic
   signaling has been noted earlier ([229]51), and genetic and
   transcriptomic analysis showed that MYC contributed to the molecular
   mechanisms underlying tumorigenesis in these murine models. CN gains in
   MYC are frequent in PTCL-GATA3 ([230]12), but the loss of TP53 also
   allows MYC expression to be up-regulated due to the disruption of
   direct TP53 regulation at the MYC promoter, which could play a role in
   breaking senescence during transformation ([231]52) rather than MYC
   being the primary driver of lymphomagenesis. Whether the MYC
   deregulation is caused by secondary genetic changes is something that
   needs to be investigated further.

   To understand the functional properties of the CD4+ T cells from the
   murine models, in vitro culture demonstrated that loss of p53 signaling
   promotes TCR-independent growth. However, Pten loss resulted in rapid
   apoptosis in mature human and murine CD4+ T cells upon TCR engagement
   but is ameliorated in the presence of Trp53 mutation under low
   serum/stress conditions, suggesting that a hyper-PI3K signal is highly
   detrimental to mature T cell survival. This may explain the observation
   of the rarity of homozygous PTEN loss in PTCL-GATA3 as PTEN expression
   levels and the associated PI3K signaling are needed to be maintained
   within a certain range for T cell growth and survival. The
   Trp53^mutPten^Δ or Trp53^ΔΔPten^Δ CD4+ T cells adapted to better
   survival under low IL-2 conditions, and this advantage disappeared when
   cultured with high IL-2. These cells may have a selective advantage
   under in vivo conditions due to their ability to grow even in limited
   IL-2 concentration, and pro-survival adaptation under stress conditions
   as observed in serum/nutrient depletion assays, thus adapting to
   changes for tumor expansion under in vivo conditions, although this was
   not evident with engraftment of preneoplastic cells from the same
   genotype. These findings with murine models were largely concordant in
   human CD4^+ T cells and suggest that alterations of several signaling
   pathways (e.g., cell cycle transition, proliferation, T cell
   activation, metabolism, and T[H]1 differentiation block), mediated by
   TP53 and PI3K deficiencies, are central events in T[H]2 cell lymphoma
   initiation. As noted, TP53 deletion facilitates T cell activation and
   survival in in vitro human CD4+ T cells.

   Whereas Pten^ΔΔTP53^Δ or Pten^ΔΔ were challenging to grow under in
   vitro conditions for a longer period, TP53^Δ and TP53^mut survived for
   long-term cell cultures and demonstrated higher TCR activity and GATA3
   expression.

   In addition, we observed that ectopic expression of TP53 in human cell
   lines down-regulated GATA3 expression. ChIP-seq and ChIP-qPCR validated
   a TP53 binding site at the intron-3 locus of GATA3 involved in the
   transcriptional regulation of GATA3, and luciferase reporter assays
   showed that inclusion of the intron-3 region in luciferase vector
   transduction reduced luciferase transcriptional activity with increased
   p53 expression compared to baseline levels. Inducing a mutation at a
   proposed “half” p53 binding motif located in this region resulted in
   increased luciferase expression, reversing the inhibition observed with
   the WT motif. The mechanistic link between TP53 and GATA3 is strongly
   evident from the ChIP-qPCR, motif enrichment analysis, and functional
   luciferase assays, thus establishing the role of p53 in regulating
   GATA3 expression.

   Together, this suggests that p53 signaling is involved in regulation of
   GATA3 and T[H]2 differentiation at this locus, beyond its canonical
   function of cell cycle regulation. Although the current understanding
   of T[H] lineage fate decisions point to extrinsic factors that drive
   naïve CD4+ T cell differentiation ([232]53, [233]54), we demonstrated
   that the intrinsic mechanisms including p53 and PI3K signaling may
   cooperate to polarize naïve CD4+ T cells toward a T[H]2 differentiation
   and block T[H]1 differentiation, as observed in polarization assays.
   The aberrations may result in the activation of STAT3 and MYC signaling
   and is a hardwired response due to PTEN and TP53 loss, consistent with
   an earlier finding that STAT3 signaling promotes T[H]2 differentiation
   ([234]33).

   We performed comparative transcriptomic analysis on the hPTCL-NOS and
   m-TCLs to establish the relevance of murine models to hPTCLs. We
   observed that PTCL-GATA3 signatures are significantly associated with
   Trp53^mutPten^Δ and Trp53^ΔΔPten^Δ m-TCLs, suggesting that murine
   models may recapitulate PTCL-GATA3 lymphomagenesis. Integrative
   analysis of m-TCLs with hPTCLs carrying del-TP53 and del-PTEN defined
   molecular distinction within PTCL-NOS (i.e., PTCL-GATA3 and
   PTCL-TBX21). The consensus mRNA signatures were enriched with a
   T[H]2-like gene expression program, PI3K, STAT3, and MYC-mediated
   transcriptional program and showed association with prognosis within
   PTCL-NOS. MYC expression is commonly increased in PTCL-GATA3 and our
   biological models. The cooperative loss of TP53 and PTEN in PTCL-GATA3
   also presents an unprecedented opportunity for therapeutic
   vulnerability by blocking PI3K activation. To demonstrate therapeutic
   relevance, we generated preclinical orthotropic models engrafted with
   Trp53^mutPten^ΔΔ or Trp53^ΔΔPten^Δ m-TCLs in NSG mice and further
   passaged in NSG mice to generate a preclinical cohort. These cohorts
   were randomized and tested a clinical-grade PI3K inhibitor (i.e.,
   duvelisib or copanlisib), and PI3K inhibitor treatment led to a
   survival advantage and reduced PI3K activation. Although the use of
   PI3K inhibition itself is not novel, we determined that its use in
   PTCL-GATA3 has a targeted value due to the dependency frequent
   up-regulation of the PI3K/AKT pathway. These data provide a rationale
   for the hard-to-target pathway (e.g., aberrant p53 signaling), which
   are vulnerable to PI3K inhibition in preclinical m-TCL models. Although
   there are no standard treatments available for PTCLs, the role of PI3K
   inhibitors can be logically extended to mature PTCLs, including
   PTCL-GATA3 or other PI3K-dependent T cell malignancies. Although PI3K
   inhibitors (e.g., duvelisib and copanlisib) are used in clinical
   settings for PTCL with T[FH] immunophenotype ([235]55, [236]56), our
   findings suggest that PTCL-GATA3 subtype with TP53 and PTEN alterations
   may show increased sensitivity to PI3K blockade. The therapeutic
   responses observed in preclinical Trp53^mutPten^ΔΔ and Trp53^ΔΔPten^Δ
   murine models supports the effectiveness of PI3K inhibitors. Although
   dysregulated TP53 and PTEN signaling represents a frequent genetic
   alteration in PTCL-GATA3, there are other targetable pathways in this
   recently identified PTCL entity. FISH and/or next-generation sequencing
   could be implemented clinically to identify TP53 and PTEN aberrations
   as our study has highlighted the prognostic importance, as well as
   identifying pathways that could be therapeutically targeted, like the
   PI3K/AKT pathway (PI3K inhibitors) or IL-2–dependent signaling
   (anti-CD25 therapy). Nevertheless, translational success may require
   additional preclinical patient-derived xenograft (PDX) models followed
   up by phase 1 clinical trials to evaluate other drug combinations for
   clinical improvement.

   In summary, our in-depth molecular and cellular characterization using
   distinct T cell murine models and innovative genetically edited human
   CD4+ T cell culture systems showed that the frequently observed TP53
   and PTEN alterations promoted GATA3 expression and T[H]2
   differentiation and activation of diverse oncogenic pathways promoted T
   cell transformation. These observations provide direction for future
   clinical-translational studies in hard-to-diagnose and hard-to-treat
   PTCL subtypes.

MATERIALS AND METHODS

   Detailed materials and methods are included in Supplementary Materials
   and Methods.

Experimental design

   The objective of the study was to generate a mouse model representative
   of PTCL-GATA3. We generated six mouse models with Trp53 and Pten
   abnormalities: Pten-heterozygous (Pten^Δ:Cd4-Cre^+/−;Pten^fl/+),
   Pten-homozygous (Pten^ΔΔ:Cd4-Cre^+/−;Pten^fl/fl), Trp53-mutant
   (Trp53^mut:Cd4-Cre^+/−;Trp53^LSL-R172H+/−), Trp53-mutant;
   Pten-heterozygous (Trp53^mut Pten^Δ: Cd4-Cre^+/−;
   Trp53^LSL-R172H+/−;Pten^fl/+), and Trp53-mutant; Pten-homozygous
   (Trp53^mut Pten^ΔΔ: Cd4-Cre^+/−;Trp53^LSL-R172H+/−; Pten^fl/fl)
   andTrp53-knockout; Pten-heterozygous (Trp53^ΔΔPten^Δ: Cd4-Cre^+/−
   Trp53^fl/fl;Pten^fl/+) and compared them to WT (Cd4-Cre^−/−) for
   phenotype and functional assessments. CRISPR-edited human CD4+ T cells
   with TP53 and PTEN abnormalities were generated to validate findings
   from murine CD4+ T cells. Preclinical evaluation of PI3Kγ/δ inhibitors
   duvelisb and copanlisib was completed using Trp53^mutPten^ΔΔ and
   Trp53^ΔΔPten^Δ allografts into NSG mice.

Patient information

   The basic clinical and pathological characteristics of PTCL-NOS cases
   are included in table S1. PTCL-NOS cases with expression of two or more
   T follicular helper (T[FH]) markers were excluded, and the remaining
   cases were subclassified into PTCL-GATA3 and PTCL-TBX21 using
   transcriptomic-based molecular diagnostic signatures ([237]5) or a
   recently defined algorithm on the basis of IHC ([238]11). The research
   protocol was approved by the UNMC and COH-MC Institutional Review Board
   under protocol number 0543-09-EP.

Human CD4+ T cell culture and genetic editing

   Human CD4+ T cells were isolated from peripheral blood mononuclear
   cells (PBMCs) of healthy donors using the EasySep Human CD4+ T cell
   Isolation Kit (STEMCELL Technologies Inc., Vancouver, Canada, #17952)
   according to the manufacturer’s instructions. Specific sequences of
   CRISPR RNA (crRNA) and genetic editing methods can be found in the
   Supplementary Materials.

Generation of murine models

   All mouse models generated were on a C57BL/6 background, bred, and
   monitored at the UNMC animal facility under the supervision of trained
   veterinarians and analyzed per the UNMC animal ethics committee
   regulations under the approved Institutional Animal Care and Use
   Committee (IACUC) protocols 16-130-12-FC and 16-131-12-FC. Specifics
   into the acquisition/breeding methods for all murine models can be
   found in Supplementary Materials and Methods. Mouse genotype analysis
   was performed by using 2 x M-PCR OPTI Mix (B4001), per the
   manufacturer’s protocol. To confirm genotypes and gene allelic
   rearrangement in the murine models, gene-specific PCR was used on
   isolated CD4+ T cells, and the detailed primers are shown in table S3.

Histological, IHC, and flow cytometry analysis of murine tumors

   Mice were monitored regularly and euthanized in accordance with IACUC
   guidelines. Necropsy was performed when indicated, and tissues were
   collected for histology, immunophenotyping, PCR, and Western blotting.
   Thymus, spleen, lymph nodes (axillary, brachial, and inguinal), and
   tumor-affected infiltrated organs of lymphoma-bearing animals were
   dissected, weighed, and fixed in 10% buffered formalin for 24 hours
   before being paraffin embedded at the Tissue Science Facility (UNMC)
   for histological analysis. Tissue sections and flow cytometry staining
   details can be found in tables S4 and S5 and Supplementary Materials
   and Methods. The cellularity of the tumors was evaluated using flow
   cytometry (ACEA NovoCyte), and data analysis was performed with
   NovoExpress software.

Murine CD4+ T cell isolation and cellular assays

   Spleens were dissected from all the genotypes of murine models and
   age-matched WT mice, and single-cell suspensions were prepared after
   red blood cell lysis and final 1x phosphate-buffered saline (PBS) wash.
   CD4+ T cells were isolated from the total splenocytes using the EasySep
   Mouse CD4+ T Cell Isolation Kit (STEMCELL Technologies). Further
   details on proliferation, apoptosis, polarization, and Western blotting
   can be found in tables S6 and S7 and the Supplementary Materials.

RNA-seq and WES analysis for murine cells or neoplasm

   RNA-seq libraries were prepared with 175 ng of RNA [RNA integrity
   number (RIN) value > 7 to 9] extracted from either CD4+ T cells or
   tumor tissue following the Illumina TruSeq RNA Sample Preparation v2
   protocol (Illumina). The samples were sequenced on a NextSeq500
   instrument to obtain 20 million, 75-bp paired-end reads. The sequencing
   reads were aligned using the STAR aligner (v2.5.3). Counts per gene
   were assessed using HTSeq (v0.9.1). HTSeq counts normalized by DESeq2
   and imported into BRB-ArrayTools for analysis. TCR and B cell receptor
   (BCR) clonotypes were analyzed using MiXCR (v3.0.7).

   For WES, libraries were captured (SureSelect XT Mouse All Exon Kit,
   Agilent), amplified, and sequenced on an Illumina sequencer as
   described earlier ([239]12). Read quality was assessed with FASTQC (v.
   0.11). High-quality reads were mapped to the mouse genome (mm10) using
   BWA-mem (v.0 0.7.17-r1188). Duplicates were marked with Picard (v.
   2.9.0), followed by base recalibration by GATK (v.4-4.1.8.1). Variant
   calling was done with Varscan2 (v2.4.4) and variant annotation with
   ANNOVAR ([240]57).

ChIP and luciferase assay

   ChIP and luciferase assay protocol specifics can be found in
   Supplementary Materials and Methods. The primer sequences for GATA3
   regions in both assays are provided in table S3. These regions were
   preselected from earlier ChIP-seq data ([241]43) as potential TP53
   binding sites. Luciferase assay was performed per the manufacturer’s
   instructions with the Dual-Luciferase Reporter Assay System (Promega
   Corp. Inc.).

Drug treatment and monitoring of subcutaneously inoculated tumor cells

   Trp53^mutPten^ΔΔ or Trp53^ΔΔPten^Δ tumor cells (1 × 10^6) were injected
   subcutaneously in NSG mice (4 to 6 weeks old), and tumor growth was
   assessed every other day. Imaging details can be found in the
   Supplementary Materials. When engrafted tumors reached a volume of ~100
   mm^3, drug treatments were started. Mice were split into two groups:
   vehicle control [10% PEG-400 (polyethylene glycol, molecular weight
   400) + 90% PBS] and duvelisib (10 mg/kg) or copanlisib (10 mg/kg).

Statistical analysis

   Comparisons between the two groups were conducted using a two-tailed
   Student’s t test. P values < 0.05 were considered significant. The data
   analysis for this manuscript was conducted using R
   ([242]https://r-project.org/). Graphs were generated using the R
   packages ggplot2 and survival and Circos ([243]http://circos.ca). The
   Kaplan-Meier method was used to estimate the OS distributions using the
   R survminer package. OS times were calculated as the time from
   diagnosis to the date of death or last contact. Patients who were alive
   at the last contact were treated as censored for the OS analysis. The
   log-rank test was used to compare survival distributions. All
   statistical tests are two sided, and P values less than 0.05 are
   statistically significant unless otherwise specified. P values were
   considered significantly different among groups, with *P < 0.05,
   **P < 0.001, and ***P < 0.0001, unless otherwise indicated in the
   figures.

Acknowledgments